Influence of insecticide resistance on the biting and resting preferences of malaria vectors in the Gambia

Background The scale-up of indoor residual spraying and long-lasting insecticidal nets, together with other interventions have considerably reduced the malaria burden in The Gambia. This study examined the biting and resting preferences of the local insecticide-resistant vector populations few years following scale-up of anti-vector interventions. Method Indoor and outdoor-resting Anopheles gambiae mosquitoes were collected between July and October 2019 from ten villages in five regions in The Gambia using pyrethrum spray collection (indoor) and prokopack aspirator from pit traps (outdoor). Polymerase chain reaction assays were performed to identify molecular species, insecticide resistance mutations, Plasmodium infection rate and host blood meal. Results A total of 844 mosquitoes were collected both indoors (421, 49.9%) and outdoors (423, 50.1%). Four main vector species were identified, including An. arabiensis (indoor: 15%, outdoor: 26%); An. coluzzii (indoor: 19%, outdoor: 6%), An. gambiae s.s. (indoor: 11%, outdoor: 16%), An. melas (indoor: 2%, outdoor: 0.1%) and hybrids of An. coluzzii-An. gambiae s.s (indoors: 3%, outdoors: 2%). A significant preference for outdoor resting was observed in An. arabiensis (Pearson X2 = 22.7, df = 4, P<0.001) and for indoor resting in An. coluzzii (Pearson X2 = 55.0, df = 4, P<0.001). Prevalence of the voltage-gated sodium channel (Vgsc)-1014S was significantly higher in the indoor-resting (allele freq. = 0.96, 95%CI: 0.78–1, P = 0.03) than outdoor-resting (allele freq. = 0.82, 95%CI: 0.76–0.87) An. arabiensis population. For An. coluzzii, the prevalence of most mutation markers was higher in the outdoor (allele freq. = 0.92, 95%CI: 0.81–0.98) than indoor-resting (allele freq. = 0.78, 95%CI: 0.56–0.86) mosquitoes. However, in An. gambiae s.s., the prevalence of Vgsc-1014F, Vgsc-1575Y and GSTe2-114T was high (allele freq. = 0.96–1), but did not vary by resting location. The overall sporozoite positivity rate was 1.3% (95% CI: 0.5–2%) in mosquito populations. Indoor-resting An. coluzzii had mainly fed on human blood while indoor-resting An. arabiensis fed on animal blood. Conclusion In this study, high levels of resistance mutations were observed that could be influencing the mosquito populations to rest indoors or outdoors. The prevalent animal-biting behaviour demonstrated in the mosquito populations suggest that larval source management could be an intervention to complement vector control in this setting.

Despite such successes, residual transmission has become increasingly spatially heterogeneous, with its intensity increasing from western to eastern Gambia, and could have been driven by specific vector population dynamics [7]. The major vector species, namely Anopheles arabiensis, An. coluzzii and An. gambiae sensu stricto (s.s.) are variably distributed throughout the country. An. arabiensis is most prevalent in the eastern Gambia while An. coluzzii and An. gambiae s.s. inhabit the western region [9,10]. However, An. arabiensis has been recently found throughout the country [11], indicating possible replacement due to successful control of other sibling species [12,13]. Moreover, the population prevalence of each vector species varies by season, whereby An. arabiensis and An. coluzzii are dominant throughout the rainy season, while An. gambiae s.s. become rarest early in the onset of dry season [9,10]. DDT and pyrethroid resistance has been reported at various degrees in all vectors, that continue to be highly susceptible to carbamates and organophosphates [11,14,15].
Host seeking and resting behavior of vectors are important metrics to evaluate the impact of control and resistance management strategies [16]. Vector behavioral adaptation, resistance selection and persistent transmission could increase during extensive scale-up of interventions, and this information can only be captured by real-time surveillance [17,18]. Hence, national malaria control programs should actively monitor behavioral dynamics in the local vector population, to inform decisions.
In the Gambia, DDT and pyrethroid resistance is widespread and associated with residual transmission [11,14]. However, the effect of control activities on vectors feeding and resting behavior remains unclear. The biting and resting preferences of An. gambiae sensu lato (s.l.) populations was investigated in the Gambia following few years of intensive vector control interventions.

Anopheles gambiae s.l. collection
Indoor and outdoor-resting adult mosquitoes were sampled from July to October 2019, during the malaria transmission season across five administrative regions in the Gambia, namely Central River Region (CRR), Lower River Region (LRR), North Bank Region (NBR), Upper River Region (URR) and West Coast Region (WCR) (Fig 1). WCR is a coastal area characterized by mangrove swamps. The remaining regions are mainly inland and have forest vegetation. Rice is mainly cultivated in CRR while cereals farming is common in all regions. Two villages were selected from each region and most of the villages are GNMCP surveillance sites with high LLIN and IRS coverage. Malaria transmission is highest in URR compared to other regions in the Gambia [7].
Indoor-resting mosquitoes were collected from sleeping rooms using pyrethrum spray collection (PSC). Twenty houses per village, at least 50m apart from each other, were randomly selected. In each village, collections were done for two consecutive days, with ten houses sampled per day. Outdoor-resting mosquitoes were sampled from pit shelter traps using prokopak aspirator. Three pit shelter traps that were 10m away from the selected compounds, were placed at different parts in each village. Both indoor and outdoor collections were conducted from 06.00 am to 09.00am in every collection day.

Mosquito identification
Morphological identification of female An. gambiae s.l. was done using identification keys as described by Gillies & Coetzee [19]. Afterwards, mosquitoes were stored individually in 96% ethanol in 1.5ml Eppendorf tube until DNA extraction. DNA was extracted separately from abdomen and head/thoraces of individual mosquitoes using Qiagen QIAxtractor robot. Species-specific genotyping PCR to identify An. arabiensis, An. melas and An. gambiae was performed using specific primers to discriminate the species as previously done [20]. This was followed by restriction enzyme digestion to specifically identify An. coluzzii, An. gambiae s.s. and their hybrids (An. coluzzii-An. gambiae s.s.) [21].

Insecticide resistance markers identification
Screening for molecular markers of target-site resistance to carbamates, DDT, pyrethroids and organophosphates was done on all samples using a probe-based assay (TaqMan SNP genotyping) [22]. The following markers were investigated: voltage-gated sodium (Vgsc)-1014F, Vgsc-1014S and Vgsc-1575Y associated with target-site mutation to DDT and pyrethroids [23][24][25]. Acetylcholine esterase (Ace)-119S, marker for carbamate and organophosphate resistance [26] and glutathione-S-transferase epsilon 2 (Gste2)-114T, involved in metabolic resistance to DDT [27] were also assayed. The TaqMan allelic discrimination assay used is a multiplex real time PCR, where primers and probes specific for each insecticide target gene were employed to discriminate susceptible (wild type) and resistant (mutant) alleles based on probe fluorescence signals [28].

Plasmodium sporozoite detection
DNA extracted from mosquito head and thoraces was used to detect sporozoites of Plasmodium falciparum, P. ovale, P. malariae and P. vivax species, employing TaqMan SNP genotyping protocol [29] which enables discriminatory identification of circum-sporozoites (CSPs) of P. falciparum from P. ovale, P. malariae and P. vivax CSPs. Genomic DNA specific to each of these Plasmodium species were analyzed in each assay as positive controls.

Blood meal identification
Extracted DNA from engorged mosquito abdomens were amplified using modified multiplex PCRs with specific primers that amplify cytochrome B genes of human and animal hosts including chicken, cow, dog, donkey, goat, horse and pig [30,31].

Statistical analyses
The proportion of each mosquito species in relation to the total number of mosquitoes captured from each region was calculated in percentage, as well as allele frequencies of indoor and outdoor-resting mosquitoes. Sporozoite positivity rate was the proportion of PCR positive mosquitoes among all mosquitoes tested. Human (HBI) and animal blood meal indices were estimated as the proportion of mosquitoes positive for human or animal hosts among those positive for all hosts. Mean differences between HBI and animal blood meal indices by vector species and resting locations were analyzed by ANOVA. Statistical analyses were done using Stata/IC 15.0 (2017 StataCorp LP).

Distribution of voltage-gated sodium channel (Vgsc) mutation markers in the vectors
Vgsc point mutations associated with DDT and pyrethroid resistance were highly prevalent and detected at varying frequencies in all vector species across all regions. Overall, An. arabiensis was found resting indoors when resistance allele frequency was higher in the indoor population, whereas An. coluzzii were resting outdoors with higher outdoor resistance. No consistent resting preference was observed in An. gambiae in the presence of mutations.
Vgsc-1014S mutation was found predominantly in indoor-resting vector populations (Table 1). In An. arabiensis, the mutation was more frequent in the indoor-resting than outdoor-resting mosquitoes regardless of the region. Vgsc-1014S was also the only mutation identified in An. gambiae s.s. and An. melas when found resting indoors.

Indoor
The carbamate and organophosphate resistance marker, acetylcholine esterase (Ace)-119S was detected only in 8 (4 indoor and 4 outdoor) An. gambiae s.s. and in one hybrid specimen in URR.

Host blood meal preference
Host blood meal origin was determined in 251 randomly selected engorged mosquito abdomens. Overall, animal and human blood meal indices were higher for indoor-than outdoorresting mosquitoes (Table 3). In all vector species, most blood meal (91%) were from animal origin. Indoor-resting An. coluzzii had the highest preference for human blood while indoorresting An. arabiensis had most preference for animal blood. Regardless of their resting location, all vector species preferred cow and donkey blood meal compared to other animals. Most vectors rarely fed on chicken and horse.

Discussion
Insecticide resistance is currently widespread among malaria vectors in the Gambia [11,14], resulting in insecticide rotation for IRS. Currently, actellic, an organophosphate insecticide is being used, whilst LLINs impregnated with deltamethrin and permethrin as recommended by WHO are distributed widely [32]. This study assessed how such vector interventions have Table 2. Sporozoite positivity rate in the eleven vector species that were infected based on their resting locations.

An. arabiensis Proportion (n)
An influenced the feeding and resting behaviour, as well as malaria transmission dynamics of the vector population. Anopheles arabiensis showed a marked preference for outdoor resting and An. coluzzii and An. melas for indoor resting. An. gambiae s.s. and An. coluzzii-An. gambiae s.s. hybrid populations were found mostly resting outdoors, but with no statistical difference from those resting indoors. Moreover, local vectors had a marked preference for animal blood. The overall sporozoite infection rate was low and infectious mosquitoes were mainly outdoorresting An. arabiensis. In this study, high frequencies of Vgsc-1014F, Vgsc-1575Y and GSTe2-114T were recorded in the malaria vector populations resting indoors and outdoors. These mutations were particularly at saturation in both indoor and outdoor-resting An. gambiae s.s. populations in URR, as well as in most of the An. coluzzii-An. gambiae s.s. hybrid samples analyzed. This magnitude of the DDT and pyrethroid-associated resistance mutations [23,24,27] recorded in the vector populations is worrying. These might have been as a result of extensive coverage of IRS and LLINs across the Gambia [7,8]. These findings suggest that effectiveness of LLINs may be compromised in the local vector populations as pyrethroids remain the only public healthapproved insecticide for LLINs [32]. Selection for resistance could have resulted from the extensive use of pyrethroids for vector control [7,8] and pest control in agriculture [33,34] in this region. Moreover, the level of resistance found in the An. coluzzii-An. gambiae s.s. hybrid population suggests an extensive gene flow between An. gambiae and An. coluzzii [35][36][37] in these settings. These observations are a concern for a possible setback to the ongoing anti-vector efforts; thus deserve close monitoring.
A possible influence of genotypic resistance on the resting behaviour of the vector populations [38], was observed from this study. A higher proportion of An. arabiensis that was found resting indoors harboured the Vgsc-1014S mutation whereas An. coluzzii exhibited an outdoor-resting behavior when all mutation markers except Vgsc-1014S were higher in the outdoor than indoor population. Anopheles gambiae s.s. with extremely high mutations were also found resting both indoors and outdoors. These indicate that resistance could be driving An. coluzzii and An. gambiae s.s. from their usual indoor resting to outdoor resting behaviour [4,39,40], a trait that could promote residual transmission [5,17,40]. Indeed, indoor resting behaviour demonstrated in the resistant vectors despite the presence of IRS and LLINs suggests that resistance could be protecting these vectors against the effect of insecticides used indoors [41,42]. This has been previously reported from studies in Kenya [43,44], where An.  gambiae s.s and An. arabiensis that had higher frequencies of Vgsc-1014F and Vgsc-1014S respectively, were predominantly found resting indoors. Similarly, association between genotypic resistance and outdoor-resting behavior was earlier found in An. coluzzii [45,46]. Any resistance-driven resting behavior in these vectors could further limit the success of anti-vector interventions currently being scaled-up in these settings [47,48]. The majority of the vector populations analyzed had a marked preference for animal than human blood meal. Anopheles arabiensis demonstrated predominant preference for animal blood meal than other vector species. This is not surprising as An. arabiensis is known to be highly zoophilic [49,50]. However, the proportion of vectors resting indoors and that have taken a blood meal either from human and animal source, could be a concern for the effectiveness of vector control measures. As animals were found only in outdoor locations in the study sites, this finding shows that the vectors took blood meal from animals outdoors and later went indoors to rest regardless of the presence of IRS and LLINs, indicating that these vectors are resistant to the insecticides being used [51]. Furthermore, the observed choice of animal blood by majority of the vectors could lead to increase in vector population that may eventually resort to biting humans in the long run and become difficult to control. Notably, alternative vector control methods such as treatment of animals with endectocides [52] and larval source management [53] could be promising tools that could be adopted by the Gambia National Malaria Control Program.
This study recorded an overall low sporozoite rate in the vector populations. Given the current low malaria prevalence in the Gambia [7], a low sporozoite rate is expected. This may reflect the impact of the scaled-up in IRS and LLINs program in the study sites which seems to successfully limit mosquito access to human blood meal indoors and consequently reducing transmission, as previously reported [54,55]. Indeed, the study sites benefitted from improved housing projects conducted in country which aimed at reducing mosquito survival as well as malaria transmission [56].
The composition of the vector species was consistent with previous studies in the Gambia where the most abundant vector was An. arabiensis, followed by An. gambiae s.s. and, An. coluzzii along with their hybrids [11,14,15]. Low density of An. melas found was as a result of our choice of villages in the West, which were not located in the coastal regions where this species breeds in salty water [57]. Remarkably, predominance of An. arabiensis could be as a result of its outdoorresting preference to avoid insecticide used in IRS and LLINs [58,59]. This leaves the highly anthropophilic and endophilic species more exposed to vector interventions, possibly leading to relative advantage that maintains the exophilic population and malaria transmission [59].

Conclusion
This study observed high levels of resistance mutations in the local vectors that could be influencing their resting behaviour. As the Gambia is in earnest preparation for pre-elimination phase of malaria, the magnitude of resistance mutations observed in the vectors in this study suggests that vectors could pose great challenges for their control using the present control paradigm. Therefore, use of larval source management as a complementary vector control measure is highly recommended.
Supporting information S1 File. Overall data including separate worksheets for genotypic resistance and species genotypes; host blood meal and sporozoite positivity data.