Abundance, diversity, and distribution of mosquito vectors in selected ecological regions of Kenya: public health implications



The diversity of mosquito arbovirus vectors was investigated to define regional risk of arbovirus transmission in Kenya. Mosquitoes were sampled between April, 2007 and December, 2010 at thirteen sites across seven administrative provinces and ecological zones. CDC light traps were used to collect mosquitoes while human-landing collection was conducted in five of the sites to target day-feeding Aedes (Stegomyia) species. Over 524,000 mosquitoes were collected and identified into 101 species, 30 of them known vectors of arboviruses endemic to Kenya. Ae. (Neomelaniconion) mcintoshi and Ae. (Aedimorphus) ochraceus were most abundant in Garissa in the arid northeastern province, and Mansonia uniformis and Mn. africana in semi-arid Baringo in the Rift Valley Province. Ae. ochraceus, Mn. africana and Mn. uniformis were also significant in Nyanza Province, while Ae. (Neomelaniconion) circumluteolus predominated in Budalangi, Western Province. Aedes (Stegomyia) aegypti was predominant in Rabai in the Coast Province but insignificant in the western and Nyanza sites. Culex pipiens was abundant in Rift Valley and Nyanza Provinces around the lake shores. This study highlights the potential for emergence and re-emergence of arboviral diseases among vulnerable populations. This calls for comprehensive mapping of vector distribution and abundance for planning focused vector control measures.


Arboviruses are principally transmitted by mosquitoes, ticks, and sand flies, causing clinical disease symptoms in humans ranging from febrile illnesses to hemorrhagic fevers. In Kenya, the first case of Rift Valley fever was reported in livestock in 1912 near Lake Naivasha in the Rift Valley Province (Montgomery and Stordy 1912, Stordy 1913). Since then, several regions have experienced incidences and/or outbreaks of many mosquito-borne arboviruses that are transmitted by different species. Arboviruses known to be circulating in different ecological zones in Kenya seem to be transmitted by different species of mosquitoes. For instance, outbreaks of Rift Valley fever have previously occurred in the North Eastern Province of Kenya in 1997/1998 (Woods et al. 2002) and in 2006/2007 (Nguku et al. 2010, Sang et al. 2010) with the later outbreak affecting other areas like Baringo in the Rift Valley Province and Kilifi in the Coast Province. During the 1997–1998 outbreak, Anopheles coustani, Mansonia africana (Theobald), and Mn. uniformis (Theobald) were implicated, although only limited entomological surveillance was conducted. However, in 2006/2007, Ae. (Neomelaniconion) mcintoshi (Theobald) and Ae. ochraceus were implicated as the principal vectors in Garissa, and Mn. uniformis and Mn. africana in Baringo in the Rift Valley. Rift Valley fever virus has also been isolated from Ae. dentatus and Ae. cumminsii (Theobald) during an inter-epidemic period at the Sukari Ranch in the Central Province and Karen in the Rift Valley Province (Linthicum et al. 1985).

During the 2006/2007 Rift Valley fever outbreak, West Nile virus was isolated from Ae. (Mucidus) sudanensis (Theobald), Pongola from Ae. mcintoshi, and Bunyamwera virus from Ae. ochraceus, among other viruses (Crabtree et al. 2009). Outbreaks of dengue-2 virus were reported along the Kenyan coast in 1982, attributed to the then-frequent trade between East Africa and India, where dengue is endemic (Myers 1971), and East Africa and the abundance of domestic and peridomestic Ae. (Stegomyia) aegypti vectors (Johnson et al. 1982). Between 2011 and 2012, an outbreak of dengue occurred in Mandera on the northern border of Somalia (unpublished data) where Ae. aegypti, a known vector of dengue, is suspected to have been involved although no entomologic survey was conducted. This species was also responsible for the 2004–2006 chikungunya outbreak in the Indian Ocean Islands (Sang et al. 2008), where 75% of the population was affected (Sergon et al. 2008). While Ae. aegypti is usually responsible for urban yellow fever outbreaks, the first reported outbreak in 1992–1993 in the Rift Valley Province of Kenya was sylvatic in nature and was associated with Ae. (Stegomyia) africanus (Theobald) and the little known Ae. (Skuse) keniensis (van Someren) (Reiter et al. 1998). In the 1940s, yellow fever antibodies were detected in human sera in most of the North Eastern Province, including Lodwar, Lokichogio, Garissa, and Marsabit (Mahaffy et al. 1946), but due to lack of accompanying entomological surveys it is unknown which vectors were involved. Recently, seroprevalence studies have shown that dengue, yellow fever, chikungunya, and West Nile viruses are circulating in Busia in western Kenya, Malindi in the Coast Province and Samburu in the Rift Valley as well as Rift Valley fever virus in Malindi and Samburu (Mease et al. 2011).

Although West Nile virus outbreaks have not been reported in Kenya, West Nile virus antibodies were detected in Marsabit and Garissa in 1968 (Henderson et al. 1970) and Ijara in 2010 (Lwande et al. 2012), North Eastern Province, a single human case was also detected in Mombasa, Coast Province, in 2005 (unpublished data). The virus is of great public health importance since its emergence in Europe and America (Gubler 2007, Petersen and Hayes 2008) and South Africa (Jupp 2001). West Nile virus was also isolated from Cx. (Culex) univittatus (Theobald) from the Rift Valley Province (Miller et al. 2000). Recent laboratory studies determined that Cx. (Culex) pipiens (Linnaeus), Cx. (Culex) univittatus (Theobald) and Cx. (Culex) vansomereni (Edwards) are competent vectors of WNV (Lutomiah et al. 2011).

Onyong ‘nyong’ was first reported in 1959/60 along the Lake Victoria basin in the Western part of Kenya where An. (Cellia) funestus (Giles) and An. gambiae (Giles) were implicated as vectors (Corbet et al. 1961). Later in 1978, a virus identical to Onyong'nyong’ was isolated from An. funestus collected from Ahero on the Kano Plain (Johnson et al. 1981). In 1994, Onyong'nyong’ was again reported in western Kenya together with chikungunya (unpublished data). In 1968, 13 Pongola virus strains were isolated from Mn. africana, Ae. (Aedimorphus) dentatus (Theobald) and Culex (Culex) zombaensis (Theobald) collected from Marsabit, while Semliki Forest virus was isolated from Anopheles funestus from the same region.

Given the long history of arbovirus transmission in Kenya, we conducted this study to determine the abundance, spatial and temporal distribution, and diversity of known arbovirus vectors of public health importance in selected regions in order to define regional risks for arbovirus transmission. Because arbovirus disease transmission is largely dependent on vector distribution and abundance, the data reported here are integral to understanding disease epidemiology, prediction and prevention, of transmission through vector mapping and targeted vector control strategies.


Study sites

Mosquitoes were sampled in 13 sites across seven administrative provinces and geographically distinct regions of Kenya (Figure 1). These sites included Kakamega, Kisumu, Busia, Isiolo, and Rabai, which were selected based on arbovirus seropositive cases arising from human febrile illness data. Kitale, Naivasha, Baringo, Garissa, and Sukari ranch were selected because they have a history of RVF outbreaks while Tana Delta and Budalangi are prone to frequent flooding, which is associated with Rift Valley fever outbreaks.

Figure 1.

Arbovirus surveillance sites.

Mosquito collection

Adult mosquitoes were collected from April, 2007 to December, 2010 for a total of 379 trap nights (five to 72 trap nights per site) using CO2-baited CDC light traps (John W. Hock Company, Gainesville, FL, U.S.A.). Additionally, human landing collection (HLC) was conducted in Rabai, Kisumu, Kakamega, Busia and Kitale to collect Ae. aegypti, since this species is a day feeder and cannot be trapped using the CDC light traps. These sites were selected for HLC because of the previously reported circulation of dengue and chikungunya viruses which are transmitted by Ae. aegypti. Sampling was done two times a year during wet seasons when mosquito population densities were expected to be high. The traps were set for 13 h from 17:00 to 06:00 every day for an average of seven consecutive days per site per visit. A total of 6,045 traps were set (60 to 1,440 traps per site) for the entire study period. HLC was conducted daily for 6 h between 12:00 and 18:00 to target day-biting mosquitoes of the Stegomyia subgenera. A total of 11,400 person-hours were spent on HLC (1,800 in Rabai and 2,400 each in Kisumu, Kakamega, Kitale, and Busia).

Mosquito identification

All mosquitoes were identified to genus or species using morphological keys, including Edwards (1941), Gillies and de Meillon (1968), Harbach (1988), and Jupp (1986). The Culex, Aedes, Mansonia, and Anopheles spp. groups consisted of mosquitoes that could not be identified to species due to specimen damage. Mosquito densities were calculated as the number of mosquitoes collected per trap per night, or bites per human per hour, for the Aedes (Stegomyia) species.


A total of 524,269 mosquitoes belonging to 11 genera and 101 species was collected between April, 2007 and December, 2010. Thirty of these species are known vectors of arboviruses endemic in Kenya. In general, the overall most abundant mosquito collection by site was recorded in Garissa 37% (199,970) followed by Baringo 31% (126,967) with Mt. Elgon giving the least collections 0.002% (831). The greatest diversity was in the genus Aedes that recorded 35 species, followed by Culex (27), Anopheles (15), Coquilettidia (7) and Eretmapodite (4). Aedomyia and Ficalbia had 3 species each, Mansonia (2), and Harpagomyia, Theobaldia and Uranotaenia (1 each). Anopheles, Culex, and Mansonia were mostly sampled in areas along the lake shores such as Baringo, Budalangi, Kisumu, and Naivasha. Ae. mcintoshi was the overall most abundant species sampled in all sites combined, 16% (90,768) followed by Cx. pipiens 13% (69,291), and Ae. (Aedimorphus) tricholabis (Edwards) 13% (66,002). The least overall sampled species, which were exclusive to Rabai at the Coast, were Ae. (Skusea) pembaensis (Theobald) (45), Eretmapodite quinquevittatus (Theobald) (40), Ae. (Aedimorphus) irritans (Theobald) (3), Ae. (Finlaya) fulgens (2), Ae. (Finlaya) longipalpis (1), Cx. (Culex) neavei (1), and Cx. (Culex) aurantapex (1).

The primary vectors of Rift Valley fever virus, Ae. mcintoshi, Ae. (Neomelaniconion) circumluteolus (Theobald), Ae. ochraceus, and Ae. sudanensis, were distributed in all sites but varied in densities. Ninety-two percent (122,388) of these species combined was collected from Garissa and only 0.004% (5) from Isiolo. Aedes mcintoshi, Ae. ochraceus, and Ae. sudanensis were almost exclusive to Garissa, 98.6%, 98.2% and 99.3%, respectively. On the other hand, most of the Ae. circumluteolus were collected from Budalangi (55.9%) and Kisumu (34.5%), while 83.2% of Ae. cumminsii (Theobald) were sampled from Naivasha and 99% of Ae. dentatus from Kitale (Table 1). A total of 38 and seven specimens belonging to the Neomelaniconion and Aedimorphus subgenera, respectively, were sampled in Baringo.

Table 1.  Densities of the primary vectors of RVFV in study sites.
 Number of mosquitoes collected per site (Average mosquito collections per trap per night) 
SpeciesBaringoaBudalangibBusiaaGarissaeIsiolodKakamegaaKisumuaKitaleaMt. ElgoncNaivashaaRabaiaSukari ranchdTana DeltacTotal
  1. Total number of traps per site –a540, b75, c60, d300, e1440.

Ae. circumluteolus 24(0)3,757(50)320(1)1(0)4(0)15(0)2,339(4)125(0)90(2)64(0)0(0)7(0)0(0)6,746
Ae. mcintoshi 14(0)1(0)2(0)86,739(60)0(0)0(0)246(0)1(0)0(0)92(0)59(0)37(0)3,608(60)90,799
Ae. tricholabis 0(0)0(0)0(0)63,778(44)0(0)0(0)0(0)469(1)0(0)1,689(3)43(0)11(0)12(0)66,002
Ae. ochraceus 1(0)61(1)111(0)33,017(23)0(0)0(0)20(0)0(0)0(0)41(0)0(0)0(0)1(0)33,252
Ae. sudanensis 1(0)6(0)0(0)2,632(2)1(0)0(0)1(0)0(0)0(0)0(0)0(0)1(0)9(0)2,651
Ae. cumminsi 5(0)4(0)1(0)879(0)0(0)403(1)5(0)273(1)0(0)7,781(14)1(0)0(0)0(0)9,352
Ae. dentatus 0(0)19(0)0(0)0(0)0(0)53(0)0(0)7,062(13)1(0)0(0)0(0)0(0)0(0)7,135
Total45(0)3,848(51)434(1)187,046(129)5(0)471(1)2,611(4)7,910 (15)91(2)9,667(17)106 (0)56 (0)3,630 (60)215,937

Mansonia species, which are also important vectors of Rift Valley fever virus, were mostly associated with lake shores of the country. Seventy-one percent of these species were sampled in Baringo and 22.8% in Kisumu. Forty-six percent of Mn. uniformis was sampled in Kisumu and 44.3% in Baringo while Mn. africana were predominant in Baringo, 83.8%, followed by Kisumu 11.6% (Table 2).

Table 2.  Densities of Mansonia spp. that are also important vectors of RVFV.
 Number of mosquitoes collected per site (Average mosquito collections per trap per night) 
SpeciesBaringoaBudalangibBusiaaGarissaeKisumuaKitaleaNaivashaaRabaiaS. ranchdTana DeltacTotal
  1. Total number of traps per site –a 540, b 75, c 60, d 300, e 1440.

Ma. africana 48,730(90)1,121(15)1(0)10(0)6,754(13)0(0)1,025(2)66(0)37(0)424(7)58,168
Ma. uniformis 16,744(31)2,137(28)27(0)6(0)17,404(32)1(0)1,315(2)49(0)40(0)105(2)37,828
Mansonia sp.10,755(20)181(2)1(0)0(0)340(1)4(0)138(0)0(0)3(0)0(0)11,422

Among members of the genus Culex collected, Cx. pipiens was overall the most predominant (42%), mostly sampled in Baringo (38.5%), while the least sampled was Cx. (Culex) bitaeniorynchus (Edwards) (1%). Culex zombaensis, a secondary vector of Rift Valley fever virus, was the second most predominant (13%) and most sampled in Naivasha (73.3%), while Cx. (Culex) antennatus (Becker), also a secondary vector of Rift Valley fever virus, was most sampled in Baringo (46.2%), followed by Naivasha (35.6%). Cx. (Culex) univittatus (Theobald), a vector of West Nile virus in most of Africa and also associated with Rift Valley fever virus, was most abundant in Kisumu (37%) and Naivasha (21.6%) (Table 3).

Table 3.  Densities of Culex species, some of which are important vectors of WNV.
 Number of mosquitoes collected per site (Average mosquito collections per trap per night) 
SpeciesBaringoaBudalangibBusiaaGarissaeIsiolodKakamegaaKisumuaKitaleaMt. ElgoncNaivashaaRabaiaSukari ranchdTana DeltacTotal
  1. Total number of traps per site –a540, b75, c60, d300, e1440.

Cx. pipiens 26,736(50)2,567(34)279(1)3,421(2)2,263(8)151(0)15,297(28)246(0)368(6)12,636(23)553(1)2,854(10)1,920(32)69,291
Cx. univittatus 1,547(3)409(5)417(1)1,264(1)1,453(5)12(0)5,029(9)39(0)41(1)2,940(5)66(0)309(1)78(1)13,604
Cx. vansomereni 22(0)5(0)144(0)467(0)25(0)162(0)350(1)245(0)0(0)6,452(12)544(1)97(0)2(0)8,515
Cx. zombaensis 3(0)28(0)10(0)47(0)72(0)641(1)417(1)59(0)11(0)15,722(29)7(0)4,424(15)1(0)21,442
Cx. antennatus 7,567(14)0(0)5(0)129(0)792(2)1(0)1,882(3)1(0)0(0)5,826(11)5(0)169(1)7(0)16,384
Cx. poicilipes 2161(4)160(2)11(0)308(0)0(0)1(0)233(0)0(0)5(0)111(0)0(0)1(0)78(1)3,069
Cx. rubinotus 2632(5)0(0)0(0)56(0)0(0)0(0)2(0)0(0)0(0)2,487(5)19(0)0(0)0(0)5,196
Cx. bitaeniorynchus 724(1)29(0)1(0)430(0)8(0)10(0)150(0)50(0)0(0)178(0)2(0)23(0)75(1)1,680
Total44,454 (83)3,534(45)2,121 (4)7,546 (4)7,060 (23)1,162 (1)28,803 (52)879 (0)625 (10)54,722 (101)1,813 (3)9,532 (33)2,717 (44)164,967

Generally the anopheline mosquitoes collected were comparatively fewer than the culicine mosquitoes in all the sites, comprising 1.5% of the total collections. The highest collection of Anopheline mosquitoes was made in Tana Delta (31.8%). Anopheles funestus was the most sampled species (42.6%), mostly in Tana Delta (69.8%), followed by An. (Anopheles) coustani (20.4%), mostly in Kisumu (41.7%). Anopheles (Anopheles) gambiae were predominant in Baringo (45.3%) and An. (Cellia) squamosus in Naivasha (83.3%) (Table 4).

Table 4.  Densities of Anopheles mosquitoes in study sites.
 Number of mosquitoes collected per site (Average mosquito collections per trap per night) 
SpeciesBaringoaBudalangibBusiaaGarissaeIsiolodKakamegaaKisumuaKitaleaMt. ElgoncNaivashaaRabaiaSukari ranchdTana DeltacTotal
  1. Total number of traps per site –a 540, b 75, c 60, d 300, e 1440.

An. coustani 3(0)126(2)11(0)347(0)6(0)6(0)670(1)2(0)17(0)327(1)32(0)22(0)39(1)1,608
An. funestus 516(1)32(0)53(0)2(0)141(0)4(0)5(0)80(0)52(1)1(0)131(0)0(0)2,345(39)3,362
An. gambiae 472(2)172(2)88(0)85(0)141(0)0(0)16(0)3(0)9(0)1(0)17(0)10(0)28(0)1,042
An. pharoensis 268(0)42(1)0(0)0(0)0(0)0(0)100(0)0(0)0(0)28(0)0(0)0(0)93(2)531
An. squamosus 10(0)71(1)4(0)0(0)1(0)2(0)13(0)0(0)0(0)549(1)0(0)0(0)9(0)659
Anopheles sp.127(0)0(0)74(0)206(0)25(0)0(0)182(0)28(0)0(0)33(0)18(0)0(0)0(0)693

Approximately 3,244 Aedes (Stegomyia) mosquitoes were collected and the collection expressed as human biting index (HBI) or bites per human per hour (b/m/h). Overall, the highest HBI were recorded in Rabai (1.4 b/m/h), followed by Kisumu (0.16 b/m/h) and Kakamega (0.08 b/m/h), while Kitale had the lowest (0.012 b/m/h). Eighty-three percent and 11.3% of the total Stegomyia sp. collection in the five sites where HLC was conducted was Ae. aegypti and Ae. simpsoni, respectively. On the other hand, Ae. metallicus was the least sampled (0.09%) and was exclusively in Rabai. Aedes aegypti was the most widespread and predominant species in all sites and constituted 84.7% of the total collection by HLC in Rabai, Kisumu (77.4%), Kitale (34.5%), Busia (45.6%) and Kakamega (91.8%). The overall highest collection of Ae. africanus was made in Kisumu (83.2%), while Ae. vittatus (78.6%) was from Rabai (Table 5).

Table 5.  Number of Aedes (Stegomyia) mosquitoes collected by HLC and the human biting rates (HBR).
 Number of mosquitoes collected per site (HBR) 
  1. a 1,800 person-hours spent on mosquito collection.

  2. b 2,400 person-hours spent on mosquito collection.

  3. HBR: Human-biting rates expressed as the number of mosquitoes biting humans per hour (b/m/h).

SpeciesCollection (HBR)Collection (HBR)Collection (HBR)Collection (HBR)Collection (HBR)Overall collections (%)
Ae. aegypti 2,185(1.2)298(0.1)10(0.004)31(0.01)168(0.07)2,692 (83)
Ae. simpsoni 356(0.2)1(0.00004)6(0.003)2(0.00008)0(0)365 (11.3)
Ae. africanus 1(0.00006)84(0.035)9(0.004)7(0.0029)0(0)101 (3.1)
Ae. vittatus 33(0.02)0(0)0(0.0)4(0.0017)5(0)42 (1.3)
Ae. metallicus 2(0.001)1(0.00004)0(0)0(0)0(0)3 (0.09)
Ae. (Stegomyia) spp.2(0.001)1(0.00004)4(0.002)24(0.01)10(0)41 (1.3)


We found that mosquito vectors of arboviruses endemic or epidemic in Kenya are distributed throughout the country, although in varying densities. In some cases, distribution of vectors was restricted to certain areas probably due to the ecological and environmental adaptation. Transmission and outbreaks of arboviruses are often associated with distinct ecological zones suited to the appropriate vector species. While the finding here suggests a strong correlation between vector distribution and arbovirus outbreaks, in some cases the range of distribution extends beyond areas associated with specific arbovirus transmission. This means that although vector distribution is an important factor in virus transmission, there are other factors that also contribute to disease distribution. These include virus presence and intrinsic and extrinsic factors (Herrera et al. 2006, Turell et al. 1985) that influence the vectorial capacity of mosquitoes and presence of amplifying vertebrate hosts. Movement of infected people, animals, or mosquitoes also introduce arboviruses into new territories. For instance, while the 1997–1998 and 2006–2007 Rift Valley fever virus outbreak affected Garissa in NE Kenya, the Kilifi outbreak was attributed to movement of infected livestock from northern Kenya (Nguku et al. 2010).

Rift Valley fever virus transmission is initiated by primary vectors collectively referred to as flood water Aedes, which appear to be restricted in distribution in Kenya. Some of the sites where primary vector species were sampled are non-endemic while others are epidemic prone. For instance while Ae. mcintoshi, Ae. ochraceus and Ae. sudanensis were predominantly found in Garissa, a Rift Valley fever hotspot area in the arid northeast part of Kenya, Ae. circumluteolus was only sampled in Budalangi and Kisumu where Rift Valley fever virus transmission has not been documented. This suggests that Ae. mcintoshi, Ae. ochraceus, and Ae. sudanensis are adapted to the arid conditions in the NE province where they are responsible for initiating and sustaining Rift Valley fever epidemics. This can be attributed to the presence of numerous and well-defined dambos which are their preferred estivation and breeding sites (Linthicum et al. 1983, 1984). In contrast, Ae. circumluteolus are adapted to the tropical rainforest climate that is characteristic of western Kenya (Budalangi and Kisumu). Apart from transmitting Rift Valley fever virus, the isolation of Pongola virus from Ae. circumluteolus in South Africa (Kokernot et al. 1957) suggests that this species can be responsible for Pongola virus transmission in the Western region, although this virus has not been associated with significant human disease (Henderson et al. 1970). Reports of past isolation of West Nile virus from Ae. sudanensis from the northeastern province of Kenya (Crabtree et al. 2009) suggest that apart from Rift Valley fever virus, this species is a potential vector of West Nile virus in the northeast region. While there has been no association of Ae. trcholabis with any arbovirus, the large numbers collected during this study indicate this species could be playing a significant, though yet-to-be known, role in arbovirus transmission and therefore needs to be closely observed.

The few Ae. mcintoshi and Ae. ochraceus sampled in Baringo correlate with the small numbers collected during the 2006–07 Rift Valley fever outbreak (Sang et al. 2010). This suggests that these species could have limited epidemiological significance in Rift Valley fever epidemics in Baringo. However, their role in maintaining the virus during interepidemic periods needs to be ascertained. Previous isolation of Rift Valley fever virus from Ae. dentatus and Ae. cumminsii (Linthicum et al. 1985) and the preference of Ae. dentatus for humans (Henderson et al. 1970) means that these species may play a significant role of Rift Valley fever virus transmission in Kitale and Naivasha where they were collected in large numbers.

The abundance of Mn. uniformis and Mn. africana in Baringo, Kisumu, and Budalangi is an indication that these species are adapted to the large swampy areas around lakes which provide preferred breeding sites. Because Rift Valley fever virus was isolated from collections of these species from Baringo during the 2006–2007 outbreak (Sang et al. 2010), it is likely that they play an important role in Rift Valley fever virus transmission. It is therefore prudent to conclude that the two RVF virus epidemic prone regions of Garissa and Baringo have different vectors that initiate and drive Rift Valley fever epidemics. This is not unusual considering that different sets of vector species of mosquitoes are known to play the role of primary and/or secondary vectors of Rift Valley fever virus in diverse ecologies in Africa. For instance, the Rift Valley fever virus outbreak in Egypt in 1977 was attributed to Cx. pipiens as the primary vector (Hoogstraal et al. 1979). In Kenya, this species seems to play a secondary role as only one pool of Cx. pipiens was found infected with Rift Valley fever virus during the 2006–2007 Rift Valley fever outbreak (Sang et al. 2010).

Apart from Mn. uniformis and Mn. africana, Ae. circumluteolus was also heavily sampled in Budalangi and Kisumu. Ae. circumluteolus is documented as being able to transmit Rift Valley fever virus (Turell et al. 1991). This means that the two regions are at high risk of Rift Valley fever virus transmission if the virus was to be introduced. This risk is more enhanced in Budalangi which is prone to flooding, a factor that is associated with Rift Valley fever outbreaks. However, lack of Rift Valley fever virus transmissions in these areas could also be attributed to lower numbers of amplifying vertebrate hosts such as livestock, incompetent primary vectors, or the general absence of the virus. Naivasha and Sukari ranch are known Rift Valley fever enzootic areas where Ae. cumminsii, Cx. zombaensis and Ae. mcintoshi, known vectors of Rift Valley fever, were sampled during the study period.

Cx. univittatus is considered to be the principal West Nile virus vector in much of Africa (Huba'lek and Halouzka 1999). West Nile virus was also previously isolated from male Cx. univittatus from the Rift valley province of Kenya (Miller et al. 2000), suggesting that it is maintained by vertical transmission in the wild. Laboratory studies have also shown that this species together with Cx. vansomereni (Edwards) are competent vectors of West Nile virus in the laboratory (Lutomiah et al. 2011). Many Cx. pipiens and Cx. univittatus mosquitoes were sampled in areas near large water bodies such as lakes. For example in Baringo where there is Lake Baringo and Bogoria, Budalangi and Kisumu (Lake Victoria) and Naivasha (Lake Naivasha). The most likely reason is that the humid conditions enhance mosquito survival, and the presence of many bird species, including migratory birds from the subtropics and Europe that are their preferred hosts, boosts the mosquito population, and therefore increases the risk of West Nile virus transmission.

Culex pipiens was also ubiquitous in distribution, the second most abundant overall, and the second most sampled in Baringo, Budalangi, Kisumu, and Naivasha. Although this species is ornithophillic, it also feeds readily on mammals including humans (Reisen and Reeves 1990, Reisen et al. 1990) hence the potential to transmit West Nile virus to humans. During the 1999 New York outbreak, Cx. p. pipiens (Linnaeus) was identified as a primary vector for West Nile virus (Turell et al. 2000), while the Kenyan strain was confirmed to be a competent vector of West Nile virus (Lutomiah et al. 2011). Several serosurveys have detected WNV circulation among humans in Kenya (Morrill et al. 1991), while West Nile virus antibodies have also been detected in bird serums collected from the Rift Valley region near Lake Bogoria and in Kisumu (unpublished data). This suggests that West Nile virus is more widespread in Kenya than it is documented. In fact omnipresence and high density of Cx. pipiens increases the risk of West Nile virus transmissions/outbreaks in most parts of Kenya including Mombasa where a human case was detected in 2005 (unpublished data). Migration of birds from one place to another may also play an important role in spreading of West Nile virus to different places especially where vectors are present.

During the 2006/2007 Rift Valley fever outbreak, Rift Valley fever virus was isolated from a pool of Cx. univittatus collected from Baringo and three pools of Cx. (Culex) bitaeniorhynchus (Giles) from Kilifi (Sang et al. 2010), suggesting that these species can significantly play a secondary role of transmitting the virus during outbreaks. Vector competence studies have also shown that Cx. zombaensis (Turell et al. 2007) and Cx. antennatus (Turell et al. 2008) populations from Kenya are competent vectors of Rift Valley fever virus. These species were collected in large numbers in Naivasha and Baringo respectively.

Ae. aegypti the principal vector of dengue virus, chikungunya, and urban yellow fever virus, as well as the Ae. simpsoni complex, predominated in the Rabai collections. The human biting index (HBI) for Ae. aegypti observed in Rabai (1.2 b/m/h) was almost seven times the combined HBIs in the other four sites (0.184 b/m/h). This suggests that circulation of dengue fever virus and chikungunya in areas such as Kisumu could probably be due to alternate vectors. Although Ae. aegypti is generally highly anthropophilic (Harrington et al. 2001), our findings suggest that the coastal Ae. aegypti population is more amenable to feeding on humans than those from other areas. This distinct difference is attributed to the existence of domestic Ae. aegypti aegypti in Rabai (Trpis and Hausermann 1975) which are known to be highly anthropophillic. Although this study did not distinguish between Ae. aegypti aegypti or formosus among the collected specimens, interbreeding between the two forms could result in hybrids with behavior that mirror both parents to some degree. It is not known whether both forms coexist in other regions of the country and what influence this has on the feeding behavior of these mosquitoes.

The anthropophilly of coastal populations may account for the current endemicity of the coast to dengue and chikungunya, explaining the recent chikungunya outbreak in Lamu (Sergon et al. 2008). Evidence that East African Ae. aegypti are among the most yellow fever competent species in the world (Tabachnick et al. 1985) also means that the Kenyan coast is consistently at higher risk of YF transmission. This is despite the recent work by Ellis et al. (2012) that showed Ae. simpsoni to be a more efficient vector of yellow fever than Ae. aegypti. But even if this is the case, Rabai still registered the highest number of Ae. simpsoni collected, which means the yellow fever risk in this region is still high. The presence of Ae. africanus in Kisumu could also easily result in sylvatic dengue spilling over to urban areas where Ae. aegypti mosquitoes are likely to be found.

In conclusion, this study has shown that arbovirus vectors are well distributed throughout Kenya in regions with previous history of outbreaks and also where transmissions have not been reported. This highlights the potential for emergence of viral diseases in vulnerable populations across the country. It has also identified high risk areas based on the densities of vectors and also shows that some species are spatially restricted, a factor that may influence disease epidemiology. Therefore, there is a need to map countrywide species distribution and abundance beyond what this study has accomplished and conduct vector competence and blood meal assays for a comprehensive assessment of arbovirus risk to public health in Kenya. This will help to institute focused vector control measures in the event of a predicted outbreak. Of great importance, though, is the need to enhance surveillance activities for important arboviruses in livestock and humans and expand the prospective entomologic studies across the country.


We thank Dunston Beti, John Gachoya, Reuben Lugalia, Daniel Ngonga, and Anthony Mutai for their expert contribution in mosquito sampling and identification. This project was fully supported by the Department of Emerging Infectious Diseases (DEID), USAMRU-K, AFHSC. The content is solely the views of the authors and does not represent the official views of AFHSC.