Aedes aegypti, the principal vector for dengue, chikungunya, and other emerging arboviruses, is closely associated with human dwellings in endemic areas, and breeds preferentially in natural and man-made container habitats of various characteristics [1–5]. In Kenya, the first dengue outbreak was reported in 1982 [6], with more recent outbreaks reported in Mombasa [7] and the northeastern town of Mandera [8]. These recent dengue outbreaks involved three different dengue virus (DENV) serotypes (DENV-1, DENV-2 and DENV-3) [7, 8]. In Kenya, dengue appears to be endemic with periodic outbreaks primarily occurring along the coast in recent years [7–9]. Earlier studies reported seroprevalence of 1.1% to 7.5% in western Kenya outside the known epidemic periods [10, 11]. Similar studies in coastal Kenya revealed DENV seroprevalence ranging from 1% to 53% [11, 12].

In Kenya, the two subspecific taxa, Ae. aegypti aegypti and Ae. aegypti formosus are the primary dengue vectors [13] with the latter being more prevalent in western Kenya [14]. Aedes ae. aegypti is more common in human dwellings, while Ae. ae. formosus, the sylvatic form, breeds in forest treeholes [13, 15]. The two subspecies were found to occur sympatrically along the coast of Kenya [13]. Larvae of Ae. aegypti and Culex spp. are often reported co-occurring in the same breeding sites [16–18]. In the coastal town of Malindi, Ae. aegypti immatures were found both indoors and outdoors at high numbers [19]. In a recent study [20], Ae. aegypti was reported to breed mostly outdoors, in what may be a novel adaptive strategy by this vector, which traditionally is considered to have adapted to the domestic environment and breed mainly in indoor water storage containers [13, 21, 22]. Predominantly outdoor breeding by Ae. aegypti has also been observed in the city of Mombasa, where recent outbreaks of dengue have been reported [23]. The outdoor environment offers more breeding opportunities due to the availability of numerous rain-filled discarded containers. Heterogeneous populations of Ae. aegypti were found to occur in such habitats for relatively long periods of time [20]. Human activities involving water storage, use and disposal of water-holding containers greatly influence Ae.aegypti breeding in individual households. Some of the key factors that may influence productivity of Ae.aegypti in different container types include, the frequency of water replenishment, the availability of food for the larvae [24], the degree of sunlight exposure [25] and container covering [26].

There are few studies on the larval habitat productivity of Ae. aegypti in Kenya [13, 14, 19], despite the evidence of dengue and chikungunya virus transmission in Kenya in recent years [7–9, 27, 28]. The presence of DENV and other arboviruses in Kenya [29], coupled with ideal ecological conditions for the vector mosquitoes, increases the risk for DENV outbreaks in this region. In the absence of a vaccine, epidemiological surveillance and vector control remain the best practices for preventing dengue outbreaks [22, 30–32]. Effective vector control depends on a good understanding of larval and adult vector ecology of which little is known in Kenya. The purpose of this study is to characterize breeding habitats and establish container productivity profiles of Ae.aegypti in rural and urban sites in western and coastal Kenya. By determining the container habitat characteristics that influence the selection of a breeding habitat by Ae.aegypti mosquitoes, key container habitats can be identified and targeted for effective larval source reduction efforts in dengue vector control programs.

In this study, we report a notable preference for outdoor breeding by Ae. aegypti, contrary to earlier findings from the region [13], but consistent with other recent findings elsewhere [20, 25, 41–43]. This is clearly indicated by significantly more Ae. aegypti immatures found in containers located outdoors, despite a higher number of containers with water indoors (Tables 1, 2). Further in support of this finding, oviposition surveys in all the study sites show significantly higher oviposition activity by Ae. aegypti outdoors than indoors. Drums, buckets, discarded tires, and small domestic containers are the key containers for Ae. aegypti development in all study sites, consistent with other studies [2, 19, 22, 44, 45]. In addition, water storage containers produced most of the immatures recorded, underscoring the importance of such category of containers in these regions. Significantly higher Ae. aegypti breeding activity was observed in the coastal region than in the west, even though the west had a notably higher number of potential breeding habitats. Container habitats of Ae. aegypti are consistently available throughout the year in all the study sites; however, Ae. aegypti was found to breed in more container types during the wet and short dry seasons.

The observed increased outdoor breeding activity by Ae. aegypti suggests an adaptation to outdoor and peridomestic habitats [20], a trend that is most likely to have epidemiologically important implications for vector control practices and prevention of virus transmission. This calls for more emphasis to be placed on the outdoor larval habitats in targeted source reduction measures in the study areas. Widespread use of insecticide impregnated bed nets in our study sites [33, 46, 47] could be a factor accounting for the preference of outdoor breeding by Ae. aegypti. Traditionally Ae. aegypti has been demonstrated to be a domestic species that is strongly endophilic occurring in bedrooms, kitchens, bathrooms and living rooms, where it rests on walls, hanging cloths, bednets and under furniture [48–50]. This may bring resting females into contact with insecticide-treated materials that are likely to kill or at least repel them from indoor resting sites. In our ongoing study (unpublished data), Aedes mosquitoes have been found to be susceptible to insecticides that are commonly used in bednets. However, further studies are required to confirm and possibly establish factors that can lead to this observed trend toward outdoor and peridomestic habitats.

Low indoor productivity in our study sites can also be attributed to human activities related to the use of domestic water receptacles. Most indoor containers are commonly used for hygiene, cooking and drinking and are subject to frequent emptying and cleaning which can effectively interrupt mosquito development. They are therefore much less likely to harbor Ae. aegypti immatures [22, 25]. Moreover, most of the indoor containers for water storage were often covered; this could have possibly contributed to many of them being unproductive. Containers with covers have been found to have a lower probability of infestation by Aedes mosquitoes [22] by preventing gravid females from accessing oviposition sites [45, 51].

Tires provided good breeding sites for Aedes mosquitoes and are responsible for producing > 30% of immatures collected from all larval habitats in outdoor locations of urban areas. The importance of tires as breeding sites of Ae. aegypti has been highlighted before [18, 19, 51–53], and recycling as a means to manage used tires in dengue control has gained popularity around the globe [30]. In the urban areas of our study sites, recycling of tires is limited to small-scale use for making sandals and soles of shoes. Therefore, intensifying other recycling options [30] is highly recommended. In addition, storage of tires in a dry environment and proper disposal of used tires should be encouraged in instances where recycling may not be feasible.

Buckets and jerrycans were found in large numbers in all the study sites, but their importance as breeding sites for Ae. aegypti mosquitoes was limited to the coastal region, especially outdoors. Unlike buckets, many jerrycans were consistently present indoors in the study sites, but they were not equally productive. Some studies have identified jerrycans as among the preferred outdoor breeding habitats [18, 19], contrary to the situation in our study sites. Low productivity by this container type can possibly be attributed to their popular usage in short-term storage of water and therefore being subject to frequent emptying and cleaning, which effectively interrupt the breeding cycle of Ae. aegypti immatures [30, 54]. Water-holding containers that are in frequent use within the domestic environment were observed to be less likely to harbor Aedes immatures [22], and this can make water storage possible without necessarily creating breeding sites for Aedes mosquitoes.

Animal feeding containers were only found in the coastal region, and their importance was particularly noticed in urban site where concrete troughs and cut tires, popular as watering points for cattle and goats in backyards of some homesteads, provided good larval sites for Ae. aegypti mosquitoes. The importance of water storage containers is due to the fact that they can hold sufficiently large volumes of water for considerably longer periods that are adequate for complete larval development. However, breeding in these containers can be eliminated by provision of tight fitting covers and mesh screens to prevent access by gravid mosquitoes [30, 45, 51].

In this study, removal or proper management of key containers could result in over a 65% reduction in Ae. aegypti immature population. This would translate into a corresponding reduction in Ae. aegypti population in the study areas, since pupae can be used as proxy estimates of adult Ae. aegypti mosquitoes [2].

Higher Ae. aegypti breeding activity at our coastal study sites can be attributed to many factors. Our coastal urban site is characterized by unplanned settlements and a large population of low income earners in the informal business sector, where poor hygiene coupled with inadequate water, sewer and waste management systems are common. These conditions have been found to contribute significantly to the proliferation of breeding sites for Ae. aegypti [55, 56]. In addition the practice of keeping domestic animals in backyards of these neighborhoods, probably to boost the low income levels, provides more breeding opportunities for the highly adaptive Ae. aegypti mosquito [57] which readily exploits the water receptacles meant for animal drinking and feeding. The comparatively higher breeding activity by Ae. aegypti observed in the coastal sites could possibly account for the dengue outbreaks reported mostly in the coastal region as observed in the recent past [9].

The year-round activity of Ae. aegypti can be associated with the lack of a reliable water supply system, and reliance on borehole water, well and rain water collection, necessitating water storage in households. During the long dry season, in particular, drums, buckets and pots become important producers of Ae. aegypti immatures in both regions. Ae. aegypti breeding in our study sites is closely related to the rainfall patterns in the two regions, with higher production of immatures during the rainy season, when environmental conditions are also optimal for adult activity, which translates into higher productivity of containers.

Of the high number of water-holding containers inspected, only a few were productive for Ae. aegypti, and most of them were located outdoors. Identification of key larval habitats for Ae. aegypti as part of vector control programs [45, 58] will help target larval source reduction measures. Dengue is primarily a problem of human activities that create breeding opportunities for potential vectors, the control of which can be achieved by physical means [30, 50]. In our study, good management practices for water storage containers [22, 50], recycling and proper disposal of discarded tires and small domestic containers [30] are recommended in order to achieve a significant reduction in Aedes population outdoors, especially during the rainy season when the latter type of containers become important breeding habitats. In addition, provision of reliable piped water supply to households in the study sites throughout the year would reduce storage of water in containers and thus control Ae. aegypti development sites.

The results of this study should be interpreted with caution since we systematically and repeatedly sampled households that were randomly selected for larval and ovitrap surveys. Though this sampling design enabled us to conduct the surveys over an extended period, we note that the study population may change its behavior over time, which in turn, may impact our findings. In addition, although our sample size of 20 sentinel houses in each of the four study sites may have been statistically inadequate, we were not able to inspect other larval sites beyond the large number in the individual sentinel households, given limited resources and larval surveys being labor intensive.

We distinguished Ae. aegypti subspecies morphologically following keys as provided by Edwards [37], Mattingly [38] and Huang [39]. While morphological identification of the subspecies is not optimal, the more sensitive molecular techniques are yet to be fully developed [59], and are not readily available in field conditions. Ae. aegypti aegypti has been reported to be predominantly domestic and peridomestic, while Ae. aegypti formosus mostly sylvatic [57, 59–61] . We identified all the Ae. aegypti in this study as Ae. aegypti aegypti. Given that all our samples are from the domestic or the immediate peridomestic environment, (within a range 10 m around sampling houses), we are confident that the presence of Ae. aegypti formosus did not have a significant impact on our findings.

The results of this study indicate that Ae. aegypti breeding habitats are abundant outdoors and are diverse both in the coastal and western regions of Kenya. However, a limited number of container types are responsible for the majority of the adult vector production. Targeting source reduction efforts toward these productive container types may be a cost-effective way to reduce the abundance of the dengue vector and arboviral transmission in these regions.