Entomological collections were conducted in Garithe village, Kilifi County along the Kenyan coast (Supplementary Figure 1). This region is highly diverse, made up of dense forests, dry thorny bushes, savannah vegetation and seasonal swamps of brackish water. There are two distinct rainy seasons: long rains which occur between April and July and short rains between October and November (21). The site was targeted for collections based on previous studies that described An. merus high density (22).
Mosquitoes were collected in 20 houses over six consecutive days during the month of November in 2019 (Supplementary Figure 1). Collections were carried out using Centers for Disease Control and Prevention light traps (CDC-LT) both indoors and outdoors for each house from dusk (1800hrs) to dawn (0600hrs), whilst coordinates were collected using eTrex® 10 (Garmin, Kansas, United States of America). The indoor traps were set in houses where at least one person spent the night during the collection period. The outdoor traps were strategically set next to the livestock shed and where livestock was absent, the trap was set approximately 5 metres from the household selected for indoor sampling. The collected mosquitoes were identified morphologically in the field laboratory(23) and sorted by physiological stage and sex. All the Anopheles spp were preserved individually in 1.5ml microcentrifuge tubes containing silica pellets and transported to the KEMRI Wellcome Trust Research Programme (KWTRP) laboratory and stored at -80˚C.
Using sterile scalpel and forceps, the female Anopheles mosquitoes were dissected and separated into distinct body parts for different assays. The legs and wings were used for Anopheles gambiae s.l sibling species identification, the head and thorax for the Plasmodium falciparum infection status, and the abdomen of blood-fed mosquitoes for trophic pattern and preference analysis as previously described(24)
Anopheles gambiae s.l sibling species speciation
Genomic deoxyribonucleic acid (DNA) was extracted from the legs and wings of mosquitoes as previously described with minor modifications (25). Briefly, the mosquito parts were transferred into 1.5 ml microcentrifuge tubes containing 50 µl of 20% chelex and crushed using polypropylene pestles. The lysate was incubated at 100˚C while shaking at 650 revolutions per minute (RPM) using a thermomixer (Eppendorf, Hamburg, Germany). The solution was then centrifuged at 10,000 x g for 2 minutes (mins) and the supernatant was transferred to a new 1.5 ml microcentrifuge tube, this was repeated twice, and the DNA was stored at -80˚C.
Anopheles gambiae s.l. sibling species were identified using a previously described method using primers that target the intergenic spacer (IGS) region of the ribosomal DNA (26). The species were distinguished by their band sizes after running agarose gel electrophoresis as follows: 153 bp for Anopheles quadrianulatus, 315 bp for Anopheles arabiensis, 390 bp for Anopheles gambiae s.s., 464 bp for Anopheles melas and 466 bp for Anopheles merus (26). An. funestus complex sibling species were also identified using PCR with primers targeting the internal transcribed Spacer Region 2 (ITS2) (27).
The abdomens of the blood-fed female Anopheles mosquitoes were crushed in 50 µl of molecular grade water using sterile polypropylene pestles. 30 µl of the lysate was mixed with 500 µl of phosphate-buffered saline (PBS) and then used to determine the source of bloodmeal using direct enzyme-linked immunosorbent assays (ELISA) as previously described(28,29) with slight modifications. The samples were tested against the anti-host immunoglobulin gamma (IgG): human, goat, bovine and chicken. The results were read visually as previously described (30).
Plasmodium falciparum sporozoite analysis
The mosquito head and thorax were crushed in 100 µl of 1X phosphate buffered saline (PBS) in 1.5 ml microcentrifuge tubes. Thereafter, 10 µl of 10X saponin was added to the lysate. The solution was then incubated at room temperature for 20 mins, and then centrifuged at 20,000 x g for 2 mins and the supernatant was discarded. The pellet was then resuspended in 100 µl 1X PBS. The solution was then centrifuged at 2000 x g for 2 mins and the supernatant was discarded. The pellet was thereafter resuspended in 50 µl of 20% chelex and DNA extracted using the procedure described above. Thereafter, the extracted DNA was used for SYBR green real-time PCR assays using primers described in Hermsen et al., 2001(31). Briefly, the RT-PCR reaction consisted of 7.5 µl of QuantiTect SYBR Green PCR master mix (Qiagen, Hilden, Germany), 0.75 µl of forward primer: 5’-GTAATTGGAATGATAGGATTTACAAGGT-3’ and 0.75µl of the reverse primer: 5’-TCAACTACGAACGTTTTAACTGCAAC-3’, 2 µl of nuclease-free water and 4 µl of the DNA. RT-PCR conditions were as follows: 95°C for 10 mins for HotStarTaq DNA Polymerase activation, 40 cycles of 95°C for 30 seconds (secs), 60°C for 45 secs, 68°C for 45 secs and finally the melt curve phase: 95°C for 15 secs, 60°C for 1 min and 95°C for 30 secs.
The highly polymorphic thioester domain (TED) region of thioester-containing protein 1 (TEP1) was amplified using the primers VB229 5’-TCAACTTGGACATCAACAAGAAG-3’ and VB004 5’- ACATCAATTTGCTCCGAGTT-3’ as previously described (19,32). Thereafter the 1088 ± 1 base pairs amplicon were cleaned using Qiaquick PCR purification kit (Qiagen, Hilden, Germany) and eluted in 15 µl of DNase free water. The samples were subjected to both Sanger and Next Generation Sequencing (NGS).
For the Sanger sequencing, the PCR amplicons were sequenced using the primers (VB004 and VB229), Big dye terminator chemistry v3.1 (Applied Biosystems, UK) and the reaction was run on an ABI 3730xl capillary sequencer (Applied Biosystems, UK). The resulting sequences chromatograms were curated, edited and aligned using CLC Main Workbench 7 (CLC Bio, Qiagen, Aarhus, Denmark).
For NGS, 75 cycles of paired-end sequencing were done on the Miseq platform using the Nextera DNA Flex library preparation protocol and Miseq reagent kit V3 (Illumina, USA). The resulting reads were demultiplexed, then FastQC v0.11.9 was used to remove indexes, low-quality PhiX and adapter reads. Identity of the resultant reads was ascertained using both BLASTn v2.11.0 and mapping onto the TEP1 references.
Multiple sequence alignment and sequence editing was conducted in AliView software v1.25, using the newly sequenced TEP1-TED sequences and those collated from GenBank(33). Maximum Likelihood phylogenies were reconstructed using the TIM+F+I substitution model with 4 gamma categories (TIM+F+I+G4) as the best fitting model as inferred by JmodelTest in iqtree v1.6.9(34) and visualized using FigTree v1.4.4.
Statistical analysis, visualisation and mapping were done using R software (v 4.1.0)(35). The data was further analysed using a generalised negative binomial regression model with a log link and robust errors whereby the dependent variable was the number of collected mosquitoes, and the independent variables were site (indicator variable), day and household using STATA v17.0 (StataCorp, College Station, TX, USA). The human blood index (HBI) was calculated as the proportion of mosquitoes that had fed on humans divided by total blood meals tested for each species
Haplotype statistics such as the number of haplotypes, haplotype frequencies and haplotype configuration and tests of natural selection: Tajima’s D(36), Fu and Li's D and Fu & Li’s F(37) were calculated using the DNA sequence polymorphism software(DnaSP)(38).