**4. Discussion**

#### *4.1. Blood-Feeding Challenges with Wild Mosquitoes*

The propensity to feed on guinea pigs was not innate to the wild mosquito population and the colony could not be maintained by other means. The tendency to blood feed on guinea pigs was introduced by out-crossing, which is evidence for the genetic basis of intrinsic host-seeking factors within this Muleba mosquito strain. Host-seeking behaviors drive host choice, which is in turn driven by adaptive advantages that result from feeding on certain host species [58–60]. Wild mosquitoes were collected from bedrooms, which could indicate a preference of these mosquitoes to human blood. Similarly, observations from other studies [61,62] have associated host preference with the availability of host species for blood-feeding, which by their abundance form a readily accessible source of blood. This plasticity in host choice within mosquitoes could also be species- or strainspecific, accounting for differences in adopting a particular host as a blood source between different mosquito species or strains, as observed in this study where wild mosquitoes had a low affinity to guinea pig blood compared to the insectary-reared Kisumu strain.

#### *4.2. Initial Low Insecticide Resistance Following Cross-Breeding*

A common method used to establish resistant mosquito strain in the insectaries involves collecting wild-resistant mosquitoes and carefully maintaining them as they adapt to insectary conditions, usually going through a narrow bottleneck of few survivors in the first few generations. However, this endeavor has its challenges, such as failure of the wild strain to adapt to insectary temperature, relative humidity, and food; reduced mating; difficulties in blood-feeding on a new blood source; and reduced insecticide resistance. Early generations (15th to 17th) of Muleba-Kis strain in this study exhibited a low level of phenotypic resistance, which could be attributed to the low frequency of resistant alleles inherited from the resistant parent, the Muleba strain. The observed low frequency of resistant alleles, due to standing variation originating from the parental line before pesticide selection, is a phenomenon reported in other studies [63].

#### *4.3. Impact of Mosquito Developmental Stage Used for Selection*

The selection at the larval stage was chosen for three reasons. First, evolutionary pressure is strongest in young individuals to increase the probability of survival to reproductive maturity. Second, beneficial mutations at an older age can be associated with harmful effects in young individuals [64–66]. Third, by exerting the selection pressure to the aquatic stage of the mosquitoes, there is assurance for successive selection as it is impossible for larvae to survive subsequent selections but only through developing resistance [67]. Additionally, many reports have associated larvae exposure to trace amounts of pesticides with the development of insecticide resistance in malaria vectors [68–71].

In another study where larvae were selected, Shidrawi observed an increase of sevenfold resistance in an *Aedes* strain with initial moderate resistance when it was selected with DDT for eight generations [50]. When Shidrawi used different insecticides for the same

strain over a different selection period, he obtained a different resistance outcome. On the other hand, in a study where adult *Anopheles* were selected [72], using a pyrethroid type II in a period of a single generation the mortality level decreased from 42% to 18% over one generation, reflecting an approximately two-fold increase in resistance. Although these results indicate that adult selection induces a more appreciable increase in resistance over a short period when compared to the larval selection, further research is needed to correlate the two stages using the same strain of mosquito and the same insecticide. Additionally, since selection in this study used different insecticides in different generations, it is difficult to determine the period without selection which is taken to reverse resistance to full susceptibility.

#### *4.4. Impact of Selection Using Pyrethroids*

The resistance of the Muleba-Kis strain was based on a cross between the field *An. gambiae* s.s. from Muleba District (fixed for L1014S mutation) and the laboratory susceptible *An. gambiae* s.s. Kisumu strain, resulting in a weak resistance in an out-crossed F1 generation.

To overcome the problem of low resistance, the selection of insect colonies using a sub-lethal concentration of insecticide has been extensively adopted to increase or induce heritable resistance [73,74]. Several studies have successfully induced resistance by selecting either adult mosquitoes [50,72,75,76] or larvae [50,77–79]. Following the insecticide selection, a pre-existing low-frequency L1014S mutation became advantageous and was selected to a higher frequency in the population. Results further indicated that out-crossing between resistant and susceptible mosquito followed by positive selection has preserved the L1014S (*kdr*-e) allele inherited from the resistant parents, as similar results were obtained in other related experiments [80]. Likewise, Song and Leu [81,82] reported the gain of rodenticide resistance alleles by susceptible house mouse *Mus musculus domesticus* through hybridization with the intrinsically resistant Algerian mouse *Mus spretus*, followed by introgression under rodenticide selection. The increased insecticide resistance and affinity to guinea pig blood observed in the Muleba-Kis strain could have been inherited via a similar mechanism and is in line with the model for the inheritance of behavioral characters in mosquitoes [83]. However, intermittent selection might be the underlying reason for the observed small rises in susceptibility of the mosquitoes, as measured by WHO susceptibility tests. This reduced resistance due to withdrawal of selection is in agreemen<sup>t</sup> with other studies [72]. Apart from maintaining selection for resistance, currently there is no utility for crossing the Muleba-Kis strain to field mosquitoes to maintain a complete genetic background to field populations, as the colony was established to serve as a close representative pyrethroid resistant strain, fixed for the L1014S mutation intended for phase-I and Phase-II studies. However, when the colony is intended for field release, such as in male sterile technique programs or when used to comprehend field population dynamics, it becomes even more important to renew the colony with field material to address the genetic drift and inbreeding effects [25,26].

#### *4.5. Differential Resistance to Type I and Type II Pyrethroids*

Pyrethroids are classified into type I and type II based on their biological responses. While type I pyrethroids result in low kill with high recovery, type II pyrethroids result in high kill with low recovery. Type I pyrethroids bind preferentially to closed channels while type II binds to open channels [84]. Research has revealed that the level of resistance in houseflies with a *super-kdr* mechanism is below 100-fold for type I and is over 200-fold for type II pyrethroids [84]. Selection of the same mosquito strain could therefore generate different resistance outcomes depending on the insecticide type, class, and concentration used, among other factors. From this study, selection of larvae with pyrethroid type I correlated with increased tolerance to type I pyrethroid papers (permethrin 0.75%) in the WHO susceptibility test, and no significant tolerance was observed against pyrethroid type II papers (alphacypermethrin, deltamethrin) following the selection. A general observation from this study indicates that type I and type II pyrethroids cause different resistance

patterns, accounting for observed mosquitoes with less sensitivity to type I pyrethroids compared to type II pyrethroids. Similar results have been observed in other studies [85]. This variation is partly attributed to the different structural conformation between type I and type II pyrethroids that affect species selectivity and pyrethroid resistance [86]. Differences in structure and biological response between type I and type II pyrethroids are therefore presumed to be the underlying reasons for the different responses to selection observed in this study.

## *4.6. Metabolic Resistance*

Although routine strain characterization by the WHO susceptibility test suggests that *kdr* was the underlying mechanism for resistance, limited PBO synergist bottle bioassay, which was done only once, indicated that mosquitoes' pre-exposure to PBO results in an increased susceptibility to permethrin by 20%, suggesting the role of metabolic resistance in this strain. However, the high susceptibility of this strain to bendiocarb and pirimiphosmethyl suggests a narrow role by metabolic resistance which requires more tests to confirm its contribution to an overall resistance. There is a need for testing for the gene expression levels, especially the CYP 450 genes which have widely been linked with metabolic resistance in malaria vectors across Sub-Saharan African [87].

#### *4.7. Intermittent Quality Control Checks and Regular Strain Authentication*

In this study, the quality of the mosquito colonies was checked to ensure that the rearing and selection procedures did not lead to contamination between strains or negative effects on the mosquito's weight or size. Underweight or undersized mosquitoes are not suitable for insecticide-testing assays, as they are more likely to be knocked down or killed by a given concentration of the insecticide. Furthermore, consistency of size is a good measure of the quality of rearing and helps to produce consistent and reproducible results provided that other rearing factors such as larval density, nutrition, environmental conditions, and microbial infection are controlled. The obtained results indicated that, despite out-crossing and insecticide selection of the strain, the weight and wing length remained fairly similar across the years, with the weight varying by only 0.001 g across four years, while wing length varied within 0.0142 mm and 0.0028 mm.

Contamination between strains held in the same facility is a regular error in mosquito rearing, especially when the same or closely related species are kept nearby [17,18,88]. The PAMVERC Test Facility keeps different strains of *An. gambiae* s.l. in different rooms and performs regular species identification using the PCR method [53] and resistance status checks to monitor for any cross-contamination. Results from characterizing the Muleba-Kisumu strain indicated that this species was identified as *An. gambiae* s.s. throughout the study, implying the absence of species contamination. *Anopheles gambiae* Muleba-Kisumu population was initially found to be partially resistant with only 30% having *kdr* fixed, but later *kdr* L1014S allele reached fixation in *A. gambiae* s.s. Muleba-Kisumu populations following the insecticide selection. These same populations exhibit strong degrees of phenotypic resistance to DDT and pyrethroid class I insecticides (permethrin).

#### *4.8. Effect of Mosquito Weight and Wing Length on Phenotypic Resistance*

Data for mosquito weight from 2016 to 2020 were normally distributed. The observed deviation in 2017 in mosquito weight could partly be attributed to changes in larvae food preparation. From 2016 to 2017, the preparation of fish flakes which are used as larvae food were microwaved at 150 degrees Celsius. However, this practice was terminated in 2018 as it was suspected to increase the nutrient content of larvae food. An increase in nutrient content or food is reported to lead to longer wings [89]. Results for median weight from 2016 to 2017 when there was no selection increased; from 2017 to 2019 weight decreased significantly; then from 2019 to 2020 the selection was ongoing and mean weight remained constant. The observed increase in weight before selection was mainly due to the nutrition regimen on the larvae. On the generations from 146–158, mosquito weight was higher, with resistance thresholds equivalent to later generations (182th to 202th) when there was relatively low but maintained weight with ongoing selection. Maintaining the mosquitoes' weight is crucial, as it is the main determinant of insecticide susceptibility, and heavier mosquitoes are more likely to survive insecticide treatment [90]. Maintaining mosquito weight from year to year is essential in getting the correct interpretation from the WHO discriminatory concentrations [90], which is fundamental in both monitoring resistance development progress and strain authentication. On the other hand, mosquito wing length results were maintained except for 2019, where they were significantly higher relative to other years. Results obtained in this study indicate that progress and status of insecticide resistance are attributed to insecticide selection and are not confounded by weight or wing length. Furthermore, in this experiment there was a detectable difference between weight and wing length, however, there were no sufficient data to prove a direct correlation between wing length and mosquito weight. Although some studies [91] have observed a correlation between weight and wing length, other studies have reported a lack of correlation between wing length and weight [54,92].
