Background
In the last decade, there have been global efforts to reduce malaria morbidity and mortality through different programmes [
1]. In Kenya, the main strategies employed by the National Malaria Control Programme have been: (1) scaling up of vector control interventions; (2) timely diagnosis and effective treatment using artemisinin-based combination therapy (ACT); and (3) intermittent preventive treatment for pregnant women (IPTP) [
2]. For vector control, the World Health Organization (WHO) recommends long-lasting insecticidal nets (LLINs), indoor residual spraying (IRS) and larval source management [
3,
4]. The most common and widespread of these methods is the use of IRS and LLINs, with LLINs being more dominant due to ease of distribution and low cost associated with their roll out [
5].
Pyrethroids remain the only class of insecticides recommended for the treatment of LLINs. This arises from their low toxicity to humans and their mode of action which entails rapid and persistent effects against mosquitoes [
6]. Until an alternative is obtained, vector control using LLINs remains heavily dependent on pyrethroids.
In Kenya, a mass bed net distribution campaign in malaria endemic areas was held in 2006, where LLINs given to pregnant women and children below the age of 5 years attained a coverage of 60% [
7]. In 2012, a near universal coverage was attained with another mass distribution of LLINs, where the goal of one net in every two people in a household was reached [
8]. This increased ITN coverage coupled to the other malaria control strategies has seen reduction in malaria transmission in many localities. Unfortunately, the sustainability of vector control using LLINs remains questionable due to insecticide resistance. Already the last decade has seen increased reports of resistance to pyrethroids in at least 27 countries in Africa [
9]. In Kenya, resistance has been reported in Western Kenya, with no report on the current state of insecticide resistance in the Kenyan Coast despite intensified vector control in the area [
10‐
12].
The main malaria vectors in the Kenyan Coast are;
Anopheles funestus sensu lato (s.l.) and
Anopheles gambiae s.l., both of which are complex species [
13]. Members of both complexes exhibit variation in their biology making it difficult to have a universal control. With LLINs targeting malaria vectors biting indoors, the occurrence of resistance could greatly impact malaria transmission dynamics. Changes in species composition, as well as, changes in their role in transmission have been reported in relation to increased bed-net use and coverage [
14,
15]. With the increased use of pyrethroids for LLINs, IRS and in agriculture mosquitoes are subjected to insecticide pressure thereby, increasing their probability of resistance [
16].
In the light of the increased bed-net coverage and the likelihood of this serving as a potential selective pressure for malaria vectors, this study sort to assess the level of phenotypic and genotypic resistance in Kwale County in the Kenyan Coast. In an effort to understand the implication of resistance on malaria transmission, species composition, sporozoite infection and indoor/outdoor proportions of malaria vectors was assessed.
Results
Species composition
A total of 1101 Anopheles adults and larvae were collected from the two villages, Marigiza (520) and Kiodomaya (581). Of these, 63.03% (n = 694) were collected as adults while 36.97% (n = 407) were collected as larvae. From the adults collected, 154 were gravid, half-gravid or blood-fed, and were placed in oviposition tubes. The oviposited eggs hatched but did not survive past the 2nd larval instar. It is worth noting that only two blood-fed An. gambiae s.l. were collected and only one laid eggs which hatched but also did not survive.
Overall, the proportion of
An. funestus s.l. collected was higher (64.40%) compared to
An. gambiae s.l. (33.97%) and other secondary malaria vectors (1.63%) (χ
2 = 650.72, df = 2, p < 0.001). The secondary malaria vectors collected include;
Anopheles squamosus (n = 7),
Anopheles coustani (n = 5),
Anopheles pharoensis (n = 5) and
Anopheles pretoriensis (n = 1) (Table
1). The proportion of
An. gambiae s.l. collected in Kidomaya was significantly higher compared to Marigiza (χ
2 = 13.10, df = 1, p = 0.003) while no differences in
An. funestus s.l. was observed between the two villages. Amongst the 374
An. gambiae s.l. collected, 88.24% were
Anopheles arabiensis, 4.81% were
An. gambiae s.s. while 6.95% did not amplify (Table
1).
Table 1
Species composition of Anopheles mosquitoes collected in Marigiza and Kidomaya villages in Kwale County, Coastal Kenya
An. gambiae
|
An. arabiensis
| 183 | 147 | 330 |
| An. gambiae s.s. | 16 | 2 | 18 |
| Not amplified | 23 | 3 | 26 |
| Total | 222 | 152 | 374 |
An. funestus
| An. funestus s.s. | 250 | 289 | 539 |
|
An. leesoni
| 24 | 1 | 25 |
|
An. parensis
| 13 | 8 | 21 |
|
An. rivulorum
| 2 | 9 | 11 |
|
An. vaneedeni
| 1 | 4 | 5 |
| Hybrids | 2 | 4 | 6 |
| Not amplified | 63 | 39 | 102 |
| Total | 355 | 354 | 709 |
An. coustani
| | 3 | 2 | 5 |
An. pharoensis
| | 0 | 5 | 5 |
An. squamosus
| | 0 | 7 | 7 |
An. pretoriensis
| | 1 | 0 | 1 |
Out of 709 An. funestus s.l. collected, PCR species identification revealed that 76.02% were An. funestus s.s., 3.53% Anopheles leesoni, 2.96% Anopheles parensis, 1.55% Anopheles rivulorum, 0.71% Anopheles vaneedeni and 0.85% hybrids, while 14.39% did not amplify. Hybrids were identified based on production of two bands corresponding to two different sibling species after PCR amplification. The composition of the hybrids was: one An. parensis/An. leesoni, one An. funestus/An. parensis and four An. vaneedeni/An. parensis. However, the hybrids identified in this study will need to be analysed further as the occurrence of hybrids has been associated with sequence similarity between other species and members of An. funestus complex in the internal transcribed spacer region 2 of the rDNA. The dominant An. funestus sibling species was An. funestus s.s. in both villages.
Outdoor and indoor collections
Overall, no difference was observed between the total numbers of mosquitoes collected outdoor and indoor. In the interest of fairly comparing indoor and outdoor proportions we considered only mosquitoes collected by light traps for analysis in this section as aspirators were not used for outdoor collections. A higher number of mosquitoes was collected outdoor (76.13%, n = 370) compared to indoor (23.78%, n = 116) (χ
2 = 132.75, df = 1, p < 0.001). Only
An. rivulorum had higher proportions indoor compared to outdoor (Table
2).
Table 2
Total number and proportion of mosquitoes collected by light trap indoors and outdoors
An. funestus s.l. | – | 397 | 24.69 | 75.31 |
| An. funestus s.s. | 280 | 22.14 | 77.86 |
|
An. leesoni
| 24 | 16.67 | 83.33 |
|
An. parensis
| 14 | 28.57 | 71.43 |
|
An. rivulorum
| 5 | 60 | 40 |
|
An. vaneedeni
| 3 | 66.67 | 33.33 |
| Hybrids | 3 | 0 | 100 |
| Not amplified | 68 | 33.82 | 66.18 |
An. gambiae s.l. | – | 74 | 22.97 | 77.03 |
|
An. arabiensis
| 55 | 23.64 | 76.36 |
| An. gambiae s.s. | 2 | 50 | 50 |
| Not amplified | 17 | 17.65 | 82.35 |
An. coustani
| – | 5 | 0 | 100 |
An. pharoensis
| – | 5 | 0 | 100 |
An. squamosus
| – | 4 | 25 | 75 |
An. pretoriensis
| – | 1 | 0 | 100 |
Insecticide susceptibility bioassay
A total of 407 F0 female adult mosquitoes aged 3–5 days old raised from larvae were used to test for susceptibility to deltamethrin and permethrin. From the 407, 72.24% (n = 294) were An. gambiae s.l., 27.03% (n = 110) An. funestus s.l. and 0.74% (n = 3) An. squamosus. Of these, 155 and 201 were exposed to deltamethrin and permethrin impregnated papers, respectively, while 51 were exposed to the control papers impregnated with silicone oil. From the 356 exposed to insecticide treated papers, an overall mortality of 76.97% was observed with a mortality rate of 75.48 and 78.11% for deltamethrin and permethrin, respectively. The mortality rate for An. gambiae s.s. Kisumu stain was 100% indicating full susceptibility to the insecticides and therefore confirming the effectiveness of the insecticide impregnated papers. No mortality was observed for the negative control with silicone oil impregnated papers, thus there was no need to correct for natural causes of mortality using the Abbott’s formula.
All
An.
funestus s.l. tested showed 100% susceptibility to both deltamethrin (n = 54) and permethrin (n = 41) (Table
3). For
An. gambiae s.l. overall mortality upon exposure to pyrethroids was 68.58% with mortality to deltamethrin and permethrin being 62.38% (n = 101) and 72.50% (n = 160), respectively both of which indicate resistance to pyrethroids. For the specific
An. gambiae s.l. species,
An. arabiensis exhibited an overall mortality to pyrethroids of 66.52% with a higher mortality to permethrin (69.93%) compared to deltamethrin (61.05%). Compared to
An. arabiensis,
An. gambiae s.s. showed a higher overall mortality to pyrethroids 92.86 with 100% mortality to permethrin and 75% mortality to deltamethrin. This indicates that
An. gambiae s.s. are susceptible to permethrin and resistant to deltamethrin. As no differences in mortality rate was observed between the two villages, Kidomaya and Marigiza mortality rate data was analysed together for both villages (Table
3).
Table 3
Mortality rate in female Anopheles mosquitoes exposed to deltamethrin and permethrin
Deltamethrin | An. funestus s.l. | – | 54 (100) |
| | An. funestus s.s. | 49 (100) |
| |
An. vaneendeni
| 1 (100) |
| | Not amplified | 4 (100) |
| An. gambiae s.l. | – | 101 (62.38) |
| |
An. arabiensis
| 95 (61.05) |
| | An. gambiae s.s. | 4 (75) |
| | Not amplified | 2 (100) |
Permethrin | An. funestus s.l. | – | 41 (100) |
| | An. funestus s.s. | 38 (100) |
| | Not amplified | 3 (100) |
| An. gambiae s.l. | – | 160 (72.50) |
| |
An. arabiensis
| 143 (69.93) |
| | An. gambiae s.s. | 12 (100) |
| | Not amplified | 5 (80) |
Knockdown resistance mutations
Three hundred
An. gambiae s.l. were genotyped for
kdr-East (L1014S) and
kdr-West (L1014F) mutations. Out of the 300 mosquitoes, 53 were from field collected adults while 247 were from the adults used for the bioassay, 79 of which had exhibited the resistance phenotype upon exposure to pyrethroid impregnated papers. Only L1014S mutation was detected in five
An.
gambiae s.s. 3 of which were homozygous while 2 were heterozygous for the L1014S allele. Of the 79 mosquitoes that had exhibited the resistance phenotype after exposure to pyrethroids only one
An. gambiae s.s. showed genotypic resistance and was heterozygous for the L1014S allele (Table
4).
Table 4
Frequency of Knockdown resistance allele in relation to phenotypes determined by WHO susceptibility bioassay in Anopheles gambiae s.s.
Resistant | 78 | 0 | 1 | 77 | 0.0064 |
Susceptible | 169 | 3 | 1 | 165 | 0.0207 |
Sporozoite infection rates
Six hundred and fifty-nine mosquitoes were tested for the presence of
P. falciparum parasites. Thirty tested positive giving an overall infection rate of 4.55%. The infection rate was higher in
An. funestus 4.94% (n = 567) compared to
An. gambiae 2.60% (n = 77), these did not differ significantly (χ
2 = 0.8364, p = 0.36). For secondary malaria vectors collected no sporozoite infection was detected. There was no difference observed between outdoor infection rate (4.35%, n = 16) and indoor infection rate (4.81%, n = 14). From the indoor collected mosquitoes, only
An. funestus s.l. (5.20%, n = 14) were infected while for outdoor collected mosquitoes both
An. funestus s.l. (4.70%, n = 1) and
An. gambiae s.l. (3.57%, n = 2) were infected. The infection rate of the specific sibling species is shown in Table
5.
Table 5
Plasmodium falciparum sporozoite infection rate of Anopheles mosquitoes from Kwale, Coastal Kenya
An. funestus s.l. | – | 567 | 4.94 |
| An. funestus s.s. | 413 | 4.84 |
|
An. leesoni
| 25 | 4.00 |
|
An. parensis
| 20 | 5.00 |
|
An. rivulorum
| 10 | 0 |
|
An. vaneedeni
| 4 | 0 |
| Hybrids | 6 | 33.33 |
| Not amplified | 89 | 3.74 |
An. gambiae s.l. | – | 75 | 2.60 |
|
An. arabiensis
| 55 | 3.51 |
| An. gambiae s.s. | 2 | 0 |
| Not amplified | 18 | 0 |
An. coustani
| – | 5 | 0 |
An. pharoensis
| – | 5 | 0 |
An. squamosus
| – | 4 | 0 |
An. pretoriensis
| – | 1 | 0 |
Discussion
The present study documents a mortality rate of 75.48 and 78.11% for deltamethrin and permethrin, respectively in malaria vectors in Kwale County. For the assessment of phenotypic resistance, WHO classifies a population into three categories based on their percentage mortality or susceptibility: A population with 100–98% mortality is regarded susceptible, 97–90% mortality indicates possible resistance that needs confirmation either using more bioassays or assessing the level of resistant genes while < 90% indicates resistance [
24]. Based on this classification, this study reveals presence of phenotypic resistance to pyrethroids in
An. arabiensis, possible resistance in
An. gambiae s.s. and susceptibility in
An. funestus s.l. In comparison to an earlier study in the Kenyan Coast [
32] that showed low level of resistance (83–93%) to deltamethrin in
An. gambiae s.l., this study shows high levels of resistance (62.38%). This increased resistance levels might be as a result of selection pressure due to increased ITN coverage. However, the contribution of agricultural insecticides should not be ignored.
Though high levels of phenotypic resistance were exhibited, the levels of kdr allele frequency were very low (1.33%). Surprisingly, while An. arabiensis was more resistant to deltamethrin and permethrin compared to An. gambiae s.s. no kdr mutation was detected in An. arabiensis suggesting that other mechanisms could be contributing to the resistance phenotype observed.
For the species composition of malaria vectors this study shows higher densities of
An. funestus s.l. in comparison to
An. gambiae s.l. This is in line with previous studies that have reported changes in species composition with relative increase in
An. funestus s.l. compared to
An. gambiae s.l. These changes have been alluded to insecticide pressure arising from the up-scaling of LTNs/LLINs and IRS [
13,
33,
34].
Although the overall density of
An. funestus s.l. was higher than that of
An. gambiae s.l, the proportion of
An. funestus s.l. reared from field collected larvae was low compared to the proportion of
An. gambiae s.l. This could have occurred as result of high mortality rate for the
An. funestus larvae in the insectary during rearing due to the difficulty associated with rearing
An. funestus [
35].
For the sibling species composition of
An. gambiae complex, our study reports
An. arabiensis as the dominant sub-species. Previously, in the Kenyan Coast and other regions in Kenya
An. gambiae s.s. was the dominant subspecies while
An. arabiensis was regarded as a secondary vector [
18]. However, since the up-scaling of vector control a reverse in the trends has been reported with a relative increase in
An. arabiensis which is regarded as a more opportunistic species relative to
An. gambiae s.s. [
13,
15].
For
An. funestus sibling species composition, this study reveals a more complex composition compared to previous studies in the area where only three subspecies were identified [
36]. The present study identifies five sibling species:
An. funestus s.s.,
An. leesoni,
An. parensis,
An. rivulorum,
An. vaneedeni and six hybrids.
An. funestus s.s. dominated the
An. funestus population with a proportion of 76.02%. This correlates with findings from other studies in the Kenyan Coast and other regions in Kenya where
An. funestus s.s. has been reported as the dominant
An. funestus sibling species [
36‐
38]. Unlike in the
An. gambiae complex, where more exophagic and exophilic species have taken over, this study shows that
An. funestus s.s. which is regarded as highly anthropophilic and endophagic remains the dominant sub-species. There is little information on the historical composition of
An. funestus sub-species in the Kenyan Coast. In most of the previous studies,
An. funestus s.l. has only been identified morphologically, thus it is difficult to tell if there has been any changes in sub-species composition over time. However, the increased complexity reported in this study compared to the two previous studies [
36,
38] could be an indicator of possible changes in species composition as a result of the current vector control strategies. However, this could change with season and needs to be evaluated for different seasons and over different years before and after introduction of insecticide-treated bed nets.
In the current study, a higher proportion of malaria vectors was collected outdoor compared to indoor. This is consistent with other studies that have reported increased proportions of malaria vectors outdoor alluded to increased ITN use [
39]. While change in species dominance has been linked to increased outdoor proportions, this might be the case for
An. gambiae s.l. due to the increase in
An. arabiensis. For
An. funestus s.l., the dominancy of
An. funestus s.s which is more endophagic could suggest a possible change in feeding and resting behaviours as a way to avoid insecticides. With the main vector control methods; LLINs and IRS targeting endophilic, endophagic and anthropophilic vectors, both changes in species composition and behavior adjustment pose a big threat to malaria control.
Contrary to recent studies in the area the current study reports a high overall
P. falciparum infection rate of 4.55% [
14,
19]. There are several plausible explanations for the high infection rate. First, increased insecticidal interventions over time might have led to reduced susceptibility of malaria vectors to insecticides used to treat nets. This means the nets become less effective in repelling, deterring and killing malaria vectors. Reduced efficacy of bed nets translates to increased human vector contact leading to high infection rate in mosquitoes [
40]. Second, differences in sampling season could lead to differences in infection rates. Higher infection rates have been reported in drier seasons compared to wet seasons [
13]. For this study mosquitoes were collected during the dry season, July to August. Third, the difference in sampling method could lead to differences in infection rates. This study used light traps and Prokopack aspirator. Light traps have been reported to increase the proportion of infected mosquitoes 2–3 times fold [
41,
42]. However, the observed high infection rate might also be as a result of the changing malaria prevalence in the Kenyan Coast. Both previous studies were conducted at a time (2009–2011) when malaria prevalence was on the fall while the present study was conducted at a time when there are reports of a rising malaria prevalence in the Kenyan Coast [
43].
Results from this study show that the rate of
P. falciparum infection outdoor was the same as that of indoor. This is consistent with other studies that have reported increased
P. falciparum infected mosquitoes outdoor with increased bed net coverage [
38]. The source of outdoor sporozoite infection remains elusive since it is still not known whether the outdoor infection is as a result of the vectors resting outdoors after an infected indoor blood meal or as a result of an infected outdoor blood meal. The latter will have dire consequences on malaria control for the reason that the current vector control methods do not control outdoor biting vectors.
The 100% insecticide susceptibility exhibited by An. funestus yet they showed the highest level of Plasmodium infection seems counterintuitive. It is possible that An. funestus acquires the infection outdoor thus having reduced contact with insecticide treated nets. However, the possibility of insecticide resistance compromising vector competence should not be ignored as this could also be a plausible explanation for the low P. falciparum infection in An. gambiae.
Authors’ contributions
DMM conceived the idea and designed the study. CWK conducted the sample collection and laboratory analysis and wrote the manuscript with active contribution from other authors. JMM was responsible for the field logistics and supervised the entomological surveys. LK, FAO and WRM offered general supervision of the study. All authors read and approved the final manuscript.