Larval habitat drives differential selection pressures on genetic insecticide resistance and metabolic enzyme plasticity in Anopheles gambiae s.l during laboratory colonization Authors: Ibrahim K Gyimah, Godwin K Amlalo, Rebecca Pwalia, Samuel S Akporh, Aaron A Lartey, Akua O Y Danquah, Dominic Acquah-Baidoo, Sampson Gbagba, Ali B I Alhassan, Joannitta Joannides, Samuel O Darkwah, Godwin A Koffa, Duncan K Athinya, Rinki Deb, Samuel K Dadzie, Jewelna Akorli Published: 2026 ABSTRACT AbstractThe evolution of insecticide resistance in mosquitoes is shaped by direct insecticide exposure and environmental conditions encountered during development. This study used an experimental evolution approach to investigate how larval habitats select for insecticide resistance profiles in An. gambiae s.l. over 10 successive generations. Mosquito larvae were collected from a single site in Accra, Ghana, and reared separately in three water types: field-collected water (FW), dechlorinated tap water (TW), and distilled water (DW) for 10 filial generations. At each generation, phenotypic susceptibility to four classes of insecticides was assessed using WHO bioassays, including synergist assays with piperonyl butoxide (PBO) and, the frequencies of the kdr-w (L995F) and ace-1 (G119S) target-site mutations were determined molecularly. Activity of detoxification enzymes, P450 monooxygenases, α- and β-carboxylesterases, and insensitive acetylcholinesterase were measured biochemically. Key selected physicochemical properties of the rearing waters were also characterized to provide contextual information on potential larval environmental influences. kdr-w mutation rapidly increased to fixation by generation F2 in mosquitoes reared in dechlorinated tap water, suggesting a strong, water-mediated selection advantage. Across the entire experiment, kdr-w mutation declined from 90–100% at F0 to ~ 63% by F10, whereas ace-1 mutation increased steadily from 60–90%. Mosquitoes reared in the nutrient and ion-rich field water consistently exhibited significantly higher detoxification enzyme activity, especially α-esterases and mixed-function oxidases, with oxidase levels elevated by as much as 32% relative to TW and DW indicating phenotypic plasticity induced by natural environmental co-factors. These findings demonstrate that larval aquatic conditions strongly influence both the genetic and biochemical foundations of insecticide resistance in adult An. gambiae s.l. Environmental co-factors within breeding habitats can shape resistance trajectories, highlighting larval source management as a potentially important strategy not only for reducing vector populations but also for mitigating the selection and persistence of resistant phenotypes.Supplementary InformationThe online version contains supplementary material available at 10.1038/s41598-026-56742-9.Keywords: Anopheles gambiae s.l., Insecticide resistance, Larval habitat, Kdr-w (L995), Ace-1(G119S), Metabolic resistance, Larval source management, Experimental evolutionSubject terms: Ecology, Ecology, Environmental sciences, Genetics, Molecular biology, Zoology Background Malaria remains a significant global health challenge, particularly in sub-Saharan Africa, where the primary vectors belong to the Anopheles gambiae sensu lato (s.l.) complex1–4 Vector control is the cornerstone of malaria prevention and relies heavily on insecticide-based interventions, namely long-lasting insecticidal nets (LLINs) and indoor residual spraying (IRS)5,6. Pyrethroids have been the principal class of insecticides used for these interventions due to their efficacy and low mammalian toxicity1,7. The widespread use of these insecticides has, however, led to the emergence and rapid spread of insecticide resistance in Anopheles populations, which now threatens the effectiveness of malaria control programs worldwide5,6,8. Mosquitoes have evolved several mechanisms to survive insecticide exposure. Among these target-site modifications and enhanced metabolic detoxification are the most-studied9,10. Target-site resistance involves genetic mutations in the protein targets of insecticides, reducing their binding affinity. A well-characterized mutation in An. gambiae s.l. include the L995F substitution (also known as kdr-w or knockdown resistance) in the voltage-gated sodium channel (vgsc) gene, which confers resistance to both pyrethroids and DDT9,11,12. Another critical mutation is the G119S substitution in the acetylcholinesterase gene (ace-1), which reduces susceptibility to organophosphates and carbamates5,8,13,14. The frequency of these mutations is a key indicator of resistance pressure in wild mosquito populations15–17. Metabolic resistance involves the upregulation of detoxification enzymes that metabolize and sequester insecticides before they can reach their intended target12,18,19. The primary enzyme families implicated are cytochrome P450 monooxygenases (P450s), glutathione S-transferases (GSTs), and carboxylesterases (α and β-esterases)20. The contribution of metabolic resistance, particularly P450s, can be assessed using the synergist piperonyl butoxide (PBO), which inhibits these enzymes and can restore susceptibility to pyrethroids in resistant mosquitoes5,18,21,22. The presence of multiple resistance mechanisms within a single mosquito population complicates vector control efforts and necessitates a deeper understanding of the evolutionary factors driving their selection and maintenance23–25. The larval environment plays a critical role in shaping the evolutionary trajectory of physiological and biochemical traits of adult mosquitoes, including their insecticide susceptibility26,27. Mosquito breeding habitats can be influenced by urban pollution and agricultural runoff, exposing immature stages of mosquitoes to a wide range of chemical contaminants, including pesticides, heavy metals, and high concentrations of organic matter6,27. Physicochemical parameters of the water, such as pH, electrical conductivity (EC), nutrient levels, and biochemical oxygen demand (BOD), do not merely influence mosquito larval development and the composition of mosquito assemblages28 but, could also act as selective filters that favour certain genotypes over others. Studies have demonstrated that larval exposure to pollutants can modulate adult longevity, reproductive fitness, and tolerance to insecticides29,30. However, there is significant lack of longitudinal data on how these habitats influence the stability, fixation, or loss of resistance alleles over time. The current study used an experimental evolution approach to investigate the impact of rearing Anopheles gambiae s.l. larvae in three distinct water types: field-collected water, dechlorinated tap water, and distilled water on resistance to insecticides. The research tracks the frequency of the kdr-w and ace-1 target-site mutations and the activity of metabolic detoxification enzymes over 10 filial generations to elucidate the influence of the larval aquatic environment on the stability or otherwise of insecticide resistance. Methods Breeding water experimental set-up Anopheles gambiae s.l. mosquito immature stages (larvae and pupae) were collected during multiple sampling events from several breeding ponds located on a vegetable farm in Opeibea (5º35′56N; 0º10′59’W), a suburb within the Accra Metropolis, Ghana. Collections were conducted between September and October 2021. In addition to collecting mosquitoes, samples of the field water were fetched into closed containers and returned to the Vestergaard-NMIMR Vector Laboratory, where immature mosquitoes were randomly divided into three breeding water types: dechlorinated tap water (TW), field-collected water (FW), and distilled water (DW) to initiate the experimental colonies. The larvae were divided into 30 rearing bowls per water type, with each bowl containing approximately 50 to 80 larvae in 200 mL of the designated water. Larvae were fed with Tetra® goldfish flakes (Tetra GmbH, Melle, Germany) until pupation when they were transferred into 30cm3 cages labelled according to the water type. Emerged adults were provided with a 10% sugar solution ad libitum. Seven to ten-day-old adult female mosquitoes were fed on sheep blood through an artificial membrane feeder (Hemotek Ltd., Blackburn, UK). An oviposition dish containing the corresponding water types was prepared and placed in each experimental cage 48–72 h after blood feeding. The eggs collected from adults emerging from the preceding generation of each experimental group were hatched in fresh aliquots of their designated breeding water type for 10 subsequent filial generations. For each of these generations designated to FW, we returned to the vegetable farm to collect field water directly from mosquito breeding ponds, for rinsing oviposition substrates to initiate egg hatching and for rearing larvae throughout development in the field water treatment. Larvae were fed as previously described. All mosquitoes were maintained under standard insectary conditions, i.e., relative humidity (RH) 75 ± 10% and temperature 27 ± 2 °C. Four independent replicates were set-up for each water type at each filial generation36. Insecticide susceptibility tests Insecticide susceptibility tests were performed with adult female An. gambiae s.l. mosquitoes at each filial generation according to World Health Organization standards (WHO)31. Representatives of the four major classes of insecticides were tested: pyrethroids (0.05% deltamethrin and 0.75% permethrin), an organophosphate (0.25% pirimiphos-methyl), an organochlorine (4% DDT), and a carbamate (0.1% bendiocarb), all applied on insecticide-impregnated filter papers (sourced from Universiti Sains Malaysia). Briefly, three replicates of 20–25 non-bloodfed female Anopheles gambiae s.l., aged 3–5 days were introduced into exposure tubes lined with insecticide-treated papers and exposed for 1 h under standard laboratory conditions. For synergist assays, mosquitoes were first exposed to 4% piperonyl butoxide (PBO) impregnated papers for 1 h, before transferring into tubes containing pyrethroid-treated papers for the 1-h exposure. All mosquitoes were transferred to holding tubes following exposure and supplied with 10% sugar solution. Mortality was recorded after 24-h. The control group for the synergist assay was exposed to only PBO without subsequent insecticide exposure and were run in parallel with the test group. This served as a quality check for the assay since PBO are not insecticides. Mortality < 5% after PBO exposure validated the assay; between 5 and 20% is corrected with the Abbott formula, and > 20% invalidates the assay. Two replicates of the silicone OP-control paper (Universiti Sains Malaysia (USM)) were used as the control for susceptibility tests and were run in parallel with the insecticide exposure. We fit a generalized linear mixed model (GLMM) to investigate insecticide, breeding water type and the filial generation as fixed effects on recorded mortality, using glmer function from the lme4 package in R59–61. The replicates were set as random effect variables. The model was fit with a binomial distribution with a logit link function since the response variable was binary (dead or alive). The odds ratios and 95% confidence intervals (CI) were calculated from the summary estimates obtained from the model. Contrast analyses of mortality per generations were performed using F0 as the reference, and significant mortality was confirmed only when p-value < 0.05 and CI does not include 1. DDT and field water (FW) were also set as the reference insecticide and water type, respectively. Screening for knockdown resistance (kdr-w L995F) and acetylcholinesterase (ace-1 G119S) gene mutations. For each water type, 30 mosquitoes from each filial generation were randomly selected from those tested (died or lived) in the WHO susceptibility assay. DNA was extracted singly from samples using 2% Cetyl trimethyl ammonium bromide (CTAB) protocol32,33. Detection of the kdr-w (L995F) and ace-1 (G119S) mutations was performed using allele-specific PCR. The Anopheles gambiae vgsc-L995F alleles were characterized by PCR with primers AGD1 [5′-ATAGATTCCCCGACCATG-3′]; AGD2 [5′-AGACAAGGATGATGAACC-3′], AGD3 [5′-AATTTGCATTACTTACGACA-3′], and AGD4 [5′-CTGTAGTGATAGGAAATTTA-3′], and the ace-1 alleles with EX3AGdir [5′-GATCGTGGACACCGTGTTCG-3′] and EX3AGrev [5′-AGGATGGCCCGCTGGAACAG-3′], using established protocols. were resolved on 2% agarose gels at 120 V for 90 min, stained with SYBR Safe and visualized under UV illumination. Genotypes were scored based on the presence or absence of allele-specific bands according to established diagnostic fragment sizes34,35,69. Positive controls of known genotype and a no‑template negative control were included in every run. A generalized linear model (GLM) was used to analyse whether observed mutation frequencies were dependent on the filial generation, mutation type (kdr-w, ace1) and water. The model was fit using a binomial distribution with a logit link function using the glm function in R, and odds ratio and 95% CI were calculated62–64. The frequencies were compared among filial generations in each water type and between water types. Linear trends of mutation frequencies from F0 to F10 were also analysed and compared63,65. The mortality and mutation frequency data were merged based on filial generation and water type to investigate correlation of the mutation frequencies with mortality observed following insecticide exposure. We performed Pearson’s product-moment correlation for each unique combination of mutation type, insecticide, and water source. The correlation coefficient (r) and the corresponding p-value were calculated for each of these combinations to determine the strength and statistical significance of the relationship66. Combinations with fewer than two data points were excluded from the correlation analysis. Biochemical assay for estimation of enzyme activity Fifty non-blood-fed, non-insecticide exposed female mosquitoes per filial generation from each breeding water type were assessed for enzyme (P450 monooxygenases (P450s), carboxylesterases (α and β-esterases), and Insensitive acetylcholinesterase (AChE)) activities36. Briefly, each mosquito was homogenized individually in 1500 µL of ice-cold 0.1 M potassium phosphate buffer (pH 7.2) using a micro-pestle. Homogenates were centrifuged at 10,000 ×g for 10 min at 4 °C, and the resulting supernatant was used immediately for enzyme assays. Biochemical assays were conducted in triplicates for each sample following standard WHO and established protocols for insecticide resistance monitoring36,67,68. For each enzyme assay, 20–100 µL of supernatant was used per well, depending on the assay protocol. Each microplate included blank wells (buffer only), homogenate from susceptible Kisumu laboratory strain (negative controls), and enzyme-specific reference standards (positive controls). All reactions were run in technical duplicate on the same plate to ensure assay reproducibility. The non-specific (α or β) esterase activity was determined with 100 µL each of mosquito homogenate, α or β naphthyl acetate substrate, and freshly prepared Fast Blue B salt to form a blue-coloured azo (blue) complex in each reaction well of a 96-well plate. The plates were incubated at room temperature for 10 min after the addition of the substrate and for an additional 2 min after the addition of Fast Blue B salt. The absorbance of β-naphthyl and α-naphthyl reactions was recorded at 540 nm and 620 nm, respectively. This allowed for the quantification of esterase activity, which reflects the hydrolytic potential of the enzyme in metabolizing ester-containing insecticides. Oxidase activity was measured as a proxy for overall cytochrome P450 monooxygenase activity using the TMBZ assay, which provides a broad estimate of mixed-function oxidase capacity. Oxidase activity was assessed with 100 µL of mosquito homogenate combined with 200 µL of tetramethylbenzidine (TMBZ) reagent and 25 µL of 3% hydrogen peroxide (H₂O₂) chromogenic reagent and incubated for 5 min at room temperature. Cytochrome C (Sigma-Aldrich, PCode: 1,002,989,384, USA) was used as positive control. Absorbance was measured at 620 nm. The degree of colour development was proportional to the level of oxidase activity, which is indicative of the mosquito’s capacity for oxidative detoxification of insecticides. Acetylcholinesterase (AChE) activity was evaluated using acetylthiocholine iodide and 5,5′-dithiobis-(2-nitrobenzoic acid) (DTNB) as reagents. 100 µL each of homogenate, ATCh/propoxur solution, and DTNB were combined in each reaction well. The assay was conducted both in the absence and presence of the carbamate inhibitor propoxur to detect insensitive enzyme variants. The absorbance was recorded immediately at time zero (T₀) and again after a 10‑minute incubation at room temperature. The rate of colour formation, resulting from the reaction between thiocholine and DTNB to produce a yellow-coloured anion of 5-thio-2-nitrobenzoic acid, was measured at 414 nm. The change in absorbance indicated the presence of altered enzyme forms associated with target site insensitivity. Total protein content in each homogenate was determined using a microplate-based protein assay. Briefly, 20 µL of the mosquito homogenate was added to each well of a 96‑well microplate, followed by 80 µL of potassium phosphate buffer and 200 µL of the protein dye reagent (Bio‑Rad). The mixture was read immediately at 620 nm using a microplate reader. Protein content for each sample was estimated using a standard curve generated from bovine serum albumin (BSA) included on each plate. The protein concentration of each sample was used to normalize enzyme activity, which was expressed as absorbance per milligram of protein. Enzyme activity profiles were compared between samples from different breeding water types from the same filial generation and across the different filial generations using one-way analyses of variance (ANOVA). When water type was detected by the ANOVA as a significant effect, we performed a Tukey’s Honest Significant Difference (HSD) post-hoc test to identify which specific water type pairs had significant differences. Additionally, to assess the overall change in enzyme activity across all generations, we performed a series of independent samples Welch’s t-tests. For each enzyme, we compared the activity levels in the FW (field water) control group to the DW (distilled water) and TW (tap water) groups separately. Physicochemical assessment of breeding water Physicochemical parameters, including biochemical oxygen demand (BOD), pH, electrical conductivity (EC), total dissolved solids (TDS), total alkalinity, chlorine, phosphate, total ammonia nitrogen, carbon dioxide, calcium, magnesium, and total hardness, were assessed in each larval habitat water type at F₀ and every other generation i.e. 5 sampling points across the 10 filial generations. Measurements were conducted using standard water quality assessment procedures. pH, EC, and TDS were measured using calibrated handheld meters, while BOD, total alkalinity, chlorine, phosphate, total ammonia nitrogen, carbon dioxide, calcium, magnesium, and total hardness were determined using standard titrimetric and colourimetric methods at the Ecological Laboratory, University of Ghana37,38. Statistical analysis was performed using Stata/IC 64 software. The Shapiro–Wilk test was applied to assess the normality of the data, and since the physiochemical parameter values of the various breeding water types did not follow a normal distribution, the Bonferroni non-parametric test was used for pairwise comparisons between water types. Results Pyrethroid-PBO overcomes persistent phenotypic resistance across all water types We investigated the An. gambiae s.l phenotypic resistance against several insecticides following continual breeding in different water types for 10 filial generations. Resistance, defined according to WHO criteria as mortality below 90%, were sustained across all generations for most insecticides, with the notable exception of pyrethroid-PBO combinations (Fig. 1). Mortalities recorded in F1- F10, except for the distilled water (DW) colony which could not survive beyond the 5th generation, were compared to the reference population (F0). The logistic regression analysis demonstrated that the baseline odds of mortality for the F0 generation in FW with DDT exposure (reference insecticide) was significantly low (OR = 0.046, 95% CI = [0.014, 0.146], p < 0.0001) (Table S1), establishing a minimal mortality rate for the founding population. Mortality in FW increased by odds of 5.23 (95%CI = [1.36, 20.12], p = 0.016) only in F8 compared to F0, assuming reference levels for all tested variables. Insecticide treatments containing the synergist piperonyl butoxide (PBO) resulted in higher odds of mortality with deltamethrin + PBO (DM + PBO) dramatically increasing the odds of death by ~ 358 (95%CI = [77.08, 1663.24], p < 0.001) compared to the reference insecticide, DDT (Table S1). While the main effects highlight the overall potency of PBO-synergized insecticides, the analysis of interaction terms revealed a more complex relationship between the factors. Significantly increased mortality was identified with the pyrethroids (PM, DM) + PBO from as early as F1 and at other specific generations (Fig. 1, Table S1) as observed for DM + PBO in tap water at F8 (p = 0.025). Therefore, among the water types compared, tap water appears to increase susceptibility to pyrethroids + PBO insecticides, though this is not consistent over 10 generations. Mortality trends of An. gambiae s.l mosquito over filial generations when bred in different water types and exposed to various insecticides. Bendiocarb (BEN), DDT, deltamethrin (DM), piperonyl butoxide (PBO), pirimiphos-methyl (PIRI), and permethrin (PM). Points represent the mean mortality from four replicates at each generation. Error bars are standard error of means. Further analysis to examine whether generational mortality trends for each insecticide differed between the breeding water types and which specific filial generations exhibited statistically significant deviations from the F₀ generation within each larval environment. Tap water (TW) reared mosquitoes showed relatively high mortality at F1 for all insecticides, except PM, PM + PBO and DM + PBO (Fig. 2). This reverted at F2 and stayed consistently similar to F0 (p > 0.05) despite observed subtle changes in mean mortality throughout the remaining generations (Fig. 2, Table S2). On the contrary, formulations containing PBO (DM + PBO, PM + PBO) were associated with significantly persistent lowered mortality from F1 to F10 regardless of water type. The odds for mortality were generally < 0.1, except F6 (OR = 7.86, 95%CI = [2.57, 24.05], p < 0.0001) where there was a one timepoint increased mortality relative to F0 in FW (Fig. 2, Table S2). Comparison of odds of mortality for each generation relative to the parent filial generation. The heatmap’s colour intensity reflects the multiplicative change in the odds of death for a given generation compared to its F0 control. An odds ratio > 2 (red) means the odds are more than twice as high as F0. A value of 1 (white) means the mortality is the same as F0 and blue squares are lower mortality than F0. Statistical significance < 0.05 is indicated as * (Table S2). Divergent evolutionary trajectories of kdr-w and ace-1 mutations across water types We fitted a generalized linear model (GLM) to estimate the probability of a mosquito possessing a specific mutation at a given generation in a particular water type and then analysed trends in the changes (slopes) mutations across generations between water types. The single nucleotide mutations L995F (kdr-w) and G119S (ace-1) of An. gambiae s.l. populations were observed across all generations (F0-F10) for each breeding water. The estimated mutation frequencies across generations revealed contrasting temporal trends for the ace-1 and kdr-w. Across all water types and generations, the kdr-w mutation occurred at significantly higher frequencies than ace-1, reflecting its substantially higher baseline frequency in the founding population rather than evidence of stronger contemporary selection. (p < 0.0001, 95%CI = [4,23, 32.99]) (Table S3). Kdr-w frequency generally fluctuated between 90 – 100%, showing a strong fixation propensity (Fig. 3A). Ace-1 mutation frequency changed from 63% at F0 in TW to 93% at F8, indicating a progressive accumulation of this mutation over time. Except in TW where kdr-w rapidly rose to fixation after a decline at F1, this mutation generally showed a declining trend from an initial value of ~ 100% at F0 in FW and DW to 76% at F3 in DW (Fig. 3B). In FW, the odds for the kdr-w mutation decreases by about 19% (OR = 0.81, 95%CI = [0.70, 0.94], p = 0.005) with each passing generation, while it increased significantly in mosquitoes bred in TW (Table S3: gen*kdr-w*TW). Mutation frequencies of insecticide resistance-associated gene across generations. The observed frequencies (A) are plotted as points at each filial generation. Estimated trends in the gene frequencies (B), obtained from GLM, compare decline or increase of the mutation across water types. FW = field water, TW = tap water, DW = distilled water. Field water composition induces significant metabolic enzyme plasticity The activity of the enzymes tested, namely; insensitive acetylcholinesterase (AChE), ⍺-esterase, β-esterase, and mixed function oxidase, varied significantly across generations (ANOVA: F = 6.669, p < 0.0001) and were influenced by water type (ANOVA: F = 4.609, p = 0.010). Considering F0-F5, to enable inclusion of DW, we observed that TW and DW often had similar activity levels with instances where either one or both differed significantly from FW. For instance, ⍺-esterase at F0 was similar for TW and DW, but at F1 all three water types differed (Fig. 4A). The differences, however, was not consistent for onward generations, but at least one water type (often FW) differed from the others. Oxidase also showed similar patterns as described for ⍺-esterase and, AChE and β-esterase was generally similar across water types (Fig. 4). Comparison of mean estimated enzyme levels in An. gambiae s.l bred in different breeding water type for 5–10 generations. The violin plots show the density distribution of the enzyme activity levels, with multi-modal peaks indicating distinct mean clusters. Different lowercase letters above bars indicate statistically significant differences among water types in the same filial generation (p < 0.05) based on post hoc multiple-comparison tests. Groups sharing the same letter (e.g., ‘a’ vs ‘a’ or ‘a’ vs ‘ab’) are not significant, whereas groups with no letters in common (e.g., “a” vs “b” or ‘ab’ vs ‘c’) differ significantly (Table S4). From F6—F10 where only FW and TW were compared, we observe similar ⍺ -esterase activity between these water types, while the other enzymes differed often at various generations. While breeding in DW appeared to reduce levels of ⍺- and β-esterases in the mosquitoes, TW did not result in significant changes in these enzymes. AChE was the least affected enzyme and was similarly reduced in DW and TW reared mosquitoes (Fig. 4B). Overall, larvae reared in nutrient-rich FW consistently exhibited significantly higher metabolic enzyme activity compared to TW and DW cohorts. This was most pronounced in oxidase levels, which were elevated by up to 32% (Fig. 5), indicating that FW induces a higher baseline detoxification capacity. The overall change in enzyme activity for DW and TW relative to FW. Left-side bar plots indicate reduced enzyme activity in DW and TW relative to FW. Red bars are significant (p < 0.05) while grey bars are not. Marked differences in the physicochemical profiles of the three water types were observed over the experimental period, providing selective landscapes for the developing larvae (Fig. 6, Fig S1). FW exhibited elevated levels of biochemical oxygen demand (BOD), electrical conductivity (EC) and total dissolved solids (TDS) peaking at 436 mg/L, 1,060 µS/cm and 805 mg/L, respectively, far exceeding the values recorded for TW and DW (p < 0.05). Nutrient concentrations were also elevated in FW compared to TW and DW. Phosphate and ammonia nitrogen (NH₄-N) levels were significantly elevated during week 1 before undergoing a significant decline in subsequent weeks (p = 0.016). Notably, these initial concentrations exceeded the recommended World Health Organization (WHO) limits58. Conversely, concentrations of calcium (Ca) and magnesium (Mg) remained consistently below the WHO guideline of 200 mg/L across all water types throughout the study period58. pH values were generally within the WHO-recommended range (6.5–8.5) for all water types (p > 0.05), though FW showed slight fluctuations between 6.91 and 7.76. Comparison of selected physicochemical parameters of different larval habitat water types used for rearing Anopheles gambiae s.l. Different lowercase letters above bars indicate statistically significant differences among groups (p < 0.05) based on post hoc multiple-comparison tests. Groups sharing the same or at least one letter (e.g., ‘a’ vs ‘ab’) are not significantly different, whereas groups different letters (e.g., “a” vs “b”) are significant. Discussion We have tracked phenotypic insecticide susceptibility, target-site mutation frequencies, and metabolic enzyme activity (Table S5) over 11 filial generations (including the founding population), to provide evidence that the larval aquatic environment plays a crucial role in the evolution of insecticide resistance profiles of adult mosquitoes. Our key findings reveal a significant decline in the frequency of the kdr-w mutation, a concurrent rise in the ace-1 mutation, and a strong induction of detoxification enzymes by nutrient and ion-rich field water. These appear to sustain the effectiveness of PBO-synergized pyrethroids, indicating a strong metabolic influence while the underlying genetic and biochemical mechanisms are dynamic and environmentally modulated. There was evidence of potentially increasing susceptibility to other classes of insecticides including non-synergized pyrethroids during continuous mosquito maintenance in dechlorinated tap water (TW) and distilled water (DW), although this was not observed to be significant given the limits of the current study. We discuss the implications of our findings for the selection of larval water sources used in laboratory colonization of Anopheles mosquitoes, demonstrating how rearing water chemistry can influence phenotypic insecticide susceptibility and the expression of resistance mechanisms in the study population. DDT was selected as a reference insecticide because it provides a sensitive and well-established phenotypic indicator of target-site resistance mediated by knockdown resistance (kdr) mutations70–72. Unlike pyrethroids, which often exhibit resistance driven by multiple interacting mechanisms, DDT resistance is largely attributable to alterations in the voltage-gated sodium channel, allowing clearer interpretation of target-site effects. Although pyrethroids are more widely used in contemporary vector control, the use of DDT as a reference facilitates mechanistic interpretation and comparison with historical resistance datasets. The most striking result of this study was the unexpected rapid increase to fixation of the kdr-w mutation by F2 in mosquito populations reared in dechlorinated tap water (TW). This pattern contrasted with field water (FW) and distilled water (DW), where mutation frequencies remained high but fluctuated. The non-monotonic fluctuations observed in both kdr-w and ace-1 frequencies in the DW colony are more plausibly explained by stochastic processes than by directional selection. Distilled water represents a chemically simplified environment that removes many potential selective agents present in field or tap water, which may reduce colony viability and amplify random allele frequency changes. Tap water, by contrast, often contains residual chlorine, heavy metals, or mineral ions that can act as xenobiotics and potentially select for individuals with more robust detoxification or stress-response systems, sometimes genetically linked to resistance loci73,74. While these findings highlight the possible role of water physicochemical factors in shaping target-site mediated pyrethroid resistance57, alternative explanations such as environmental fluctuations must also be considered. In contrast to kdr-w, the frequency of ace-1 mutation, which confers resistance to organophosphates (OPs) and carbamates, remained lower and never reached fixation in any water type. Although ace-1 resistance alleles are often associated with fitness costs in the absence of insecticide pressure70, our results indicate that ace-1 frequency increased during the early filial generations (F₁-F₅) under distilled and tap water conditions, suggesting a relaxation or buffering of such costs in simplified larval environments. In contrast, a reduction in ace-1 frequency was observed under field water conditions and in later generations, consistent with environment-dependent expression of fitness trade-offs39,40. The generational decline in kdr-w frequencies supports the hypothesis of a fitness cost, a phenomenon described in other vector populations where resistance alleles decrease in frequency when selection pressure is removed41. For instance, a significant fitness cost associated with the L995S-kdr mutation in a Ugandan An. gambiae population led to a restoration of susceptibility in an insecticide-free environment41. Our study provides further evidence for this phenomenon with the L995F allele during laboratory colonization. These results highlight that in the maintenance of mosquito populations in the laboratory the breeding water source should be carefully monitored, especially when the mosquitoes will be used for investigating mechanisms of insecticide resistance. Even without active insecticide application, the genetic architecture of resistance within a population can evolve rapidly during colonization, potentially predisposing it to resistance against different insecticide classes. While the observed generational changes in ace-1 frequency are consistent with differential selective pressures associated with larval habitat conditions, caution is warranted in interpreting these patterns as evidence of adaptive evolutionary change over the relatively short duration of the experiment. In addition to selection, founder effect may have contributed to the resistance dynamics observed. At colony establishment, random sampling of individuals into the different larval habitat treatments may have introduced founder effects that influenced initial allele frequencies. Importantly, founder effect may also have acted throughout the experiment, as colony sizes, sex ratios and population abundance were not explicitly controlled or quantified at each stage. Furthermore, variation in larval crowding and the removal of female mosquitoes for insecticide susceptibility assays, may have altered effective population sizes and contributed to random fluctuations in ace-1 frequencies across generations. Consequently, the observed resistance trajectories are best interpreted as reflecting the combined effects of selection and stochastic demographic processes rather than selection alone. Our enzyme analyses showed dramatic effects of water type on detoxification capacity. Mosquitoes reared in field water, which was characterized by a high organic and ionic load, consistently showed elevated levels of detoxification enzymes, particularly ⍺-esterases and mixed-function oxidases (P450s). Oxidase activity showed the largest reduction of 32.1% in DW relative to FW. Similarly, ⍺-esterase activity significantly decreased in DW but was stable in TW. Oxidases and ⍺-esterases are crucial for metabolic resistance against multiple insecticide classe42, hence their reduced activities in the artificial waters (DW, and to a lesser extent, TW) suggests that natural inducers present in the FW environment are essential for maintaining the high baseline levels of these key metabolic enzymes. Their absence in the laboratory waters may effectively reduce the mosquitoes’ intrinsic detoxification capacity. β-esterases confer metabolic resistance by sequestering insecticides such as organophosphates, carbamates, and pyrethroids before these compounds reach their neural targets. Elevated β-esterase activity, often resulting from gene overexpression, enhances the breakdown of insecticides and consequently reduces their toxicity to mosquitoe43–46. Acetylcholinesterase (AChE), on the other hand, is an essential enzyme that hydrolyzes the neurotransmitter acetylcholine to terminate synaptic transmission. Organophosphate and carbamate insecticides specifically target AChE by irreversibly or reversibly inhibiting its activity, leading to the accumulation of acetylcholine at synaptic junctions, neural overstimulation, and ultimately, insect death45,47,48. All the laboratory water types (DW and TW) showed a significant decrease for β-esterase and AChE activity across generations. This suggests a mechanism of phenotypic plasticity, where exposure to environmental xenobiotics in the larval stage “primes” the adult’s detoxification systems, potentially conferring enhanced tolerance to insecticides later in life. Our findings align with a growing body of evidence highlighting the link between the larval environment and adult insecticide susceptibility26,49. Pollution and agricultural runoff in breeding sites contribute to increased insecticide resistance in Anopheles vectors49,50. The physicochemical analysis of our field water, which revealed high levels of biochemical oxygen demand (BOD), electrical conductivity (EC), and nutrients, corresponds with characteristics of polluted and agricultural habitats known to select for resistance38,49,51. The central role of metabolic resistance, particularly mediated by P450 enzymes, is well-documented across Africa19,52,53. The restoration of susceptibility with PBO in our study is consistent with findings from Benin, Cameroon, and Nigeria, where P450s like CYP6M2 and CYP6P3 are major drivers of pyrethroid resistanc52–55. The profound increase in mortality when mosquitoes were exposed to deltamethrin and permethrin in combination with the synergist PBO strongly implicates mixed-function oxidase enzymes as a primary mechanism of metabolic resistance in this population. The odds of death for deltamethrin with PBO were approximately 358 times higher than the baseline, confirming that inhibiting these enzymes can largely restore pyrethroid susceptibility. However, the observation that the efficacy of these synergized insecticides diminished in later generations suggests that the mosquito colony was adapting, possibly through the selection of alternative resistance mechanisms not fully inhibited by PBO. The insecticide resistance mechanisms in the mosquito population studied suggested by our results is that the complex chemical matrix of natural breeding sites induces a broad, non-specific upregulation of the mosquito’s detoxification machinery. The higher organic and nutrient content in field water may contribute to environmental conditions associated with a more robust enzymatic defense system, although this study does not isolate these factors as direct causal drivers 20. This has critical implications for vector control. It suggests that larval source management (LSM) is not merely a tool for population reduction but could also be a vital component of insecticide resistance management. By cleaning or modifying polluted breeding sites, control programs could potentially reduce the selection for metabolically resistant mosquitoes, thereby producing adult populations that are more susceptible to existing insecticides56. The observed shift from kdr-w based resistance towards ace-1 dominance in the laboratory colony underscores the evolutionary plasticity of An. gambiae. This implies that if vector control programs were to switch from pyrethroids to organophosphates or carbamates to manage kdr-w resistance, this population already possesses a high frequency of the corresponding ace-1 resistance allele, which could lead to rapid control failure. This highlights the necessity for integrated resistance management strategies that include rotational insecticide use, synergists, and non-chemical interventions like LSM. This study has several notable strengths. The longitudinal design, tracking a single mosquito colony for 10 filial generations, allowed for the observation of micro-evolutionary changes in resistance mechanisms over time, providing insights that cross-sectional studies cannot offer. The multi-pronged approach, integrating phenotypic bioassays, molecular genotyping, and biochemical enzyme analysis, provides a holistic and robust characterization of the resistance profile. Finally, the use of three distinct and well-characterized water types enabled a controlled comparison to isolate and understand the specific impact of breeding water quality on resistance dynamics. Despite these strengths, the study has limitations. First, laboratory colonization inherently differs from natural conditions. Factors such as predation, competition, and fluctuating environmental conditions are absent, which may alter the fitness costs and benefits of resistance alleles. Therefore, the observed decline in kdr-w frequency might be more pronounced in the laboratory than it would be in the field, where pyrethroid pressure persists. Second, the mosquito population originated from a single urban site in Accra, and the findings may not be generalizable to other An. gambiae populations in different ecological settings, such as rural agricultural areas. Insecticide residues present in field-collected breeding water were not chemically quantified. Therefore, no direct causal inference can be made regarding the contribution of specific minerals, physicochemical characteristics, or unmeasured insecticide residues to the resistance trends observed in this study. These findings should be interpreted as evidence of association within a complex larval environment rather than proof of a single primary driver. Field water may contain trace levels of agricultural or domestic insecticides that could influence resistance dynamics during larval development. However, no insecticides were deliberately introduced into any rearing water, and the observed patterns likely reflect a combination of larval environmental complexity and differential expression of fitness costs associated with resistance mechanisms. Future studies incorporating chemical characterization of breeding waters would help to further disentangle the relative contribution of environmental contaminants to resistance trajectories. This study focused on two well-established target-site resistance markers, kdr-w (L995F) and ace1 (G119S), selected based on their direct relevance to the insecticide classes evaluated and their well-characterized functional roles in Anopheles mosquitoes. While other resistance-associated markers, including metabolic variants such as GSTe2 I114T or CYP6P3 E205D, may also contribute to resistance phenotypes, their inclusion was beyond the scope of the present work. As such, the molecular analyses presented here capture major target-site mechanisms but do not represent an exhaustive survey of all possible resistance pathways. Future studies integrating broader metabolic and genomic markers would provide a more comprehensive understanding of resistance evolution under contrasting larval environments. Similarly, while the increased enzyme activity in field water reared mosquitoes is likely due to chemical induction, it cannot be entirely disconnected from potential nutritional benefits, as the rich microbial community and dissolved organic matter might produce healthier, larger adults with inherently more active metabolic systems. Lastly, the collapse of the distilled water colony after the fifth generation limited long-term comparisons for this colony. This study opens several avenues for future research. Transcriptomic (RNA-seq) analyses should be conducted on larvae and adults reared in different water types to identify upregulated detoxification genes more comprehensively. This would provide a more precise understanding than the broad enzyme activity assays used here. To confirm the fitness cost hypothesis, targeted studies measuring life-history traits (e.g., longevity, fecundity, mating competitiveness) should be performed on mosquitoes with different kdr-w and ace-1 genotypes from this population. Furthermore, replicating this longitudinal study with populations from different ecological zones, particularly rural agricultural areas with high pesticide use versus populations from fields without agricultural activities, would be crucial to determine the broader applicability of these findings. Finally, semi-field studies are needed to validate whether the trends observed in the laboratory, especially the decline of kdr-w, hold under more natural environmental conditions. Conclusions This research demonstrates that the larval aquatic environment is a powerful selective force that significantly shapes the genetic and biochemical basis of insecticide resistance in adult Anopheles gambiae s.l.. The quality of breeding water not only induces metabolic detoxification systems that can enhance insecticide tolerance but also influences the evolutionary trajectory of key target-site mutations over generations. These findings underscore that a comprehensive approach to malaria vector control must extend beyond targeting adult mosquitoes and incorporate larval source management as a strategic tool for mitigating the evolution and spread of insecticide resistance. The chemical ecology of mosquito breeding sites is a critical, and often overlooked, battleground in the fight against malaria. Supplementary Information Below is the link to the electronic supplementary material. Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Supplementary Material 5 Supplementary Material 6 Acknowledgements The authors are grateful to staff of the Ecological Laboratory, University of Ghana, for water sample analyses. The support and insightful contributions from the faculty of the African Regional Postgraduate Programme in Insect Science (ARPPIS), University of Ghana, are also gratefully acknowledged. List of abbreviations Acetylcholinesterase Biochemical oxygen demand Confidence interval Cetyl trimethyl ammonium bromide Dichlorodiphenyltrichloroethane Deltamethrin 5,5′-Dithiobis-(2-nitrobenzoic acid) Distilled water Electrical conductivity Field water Generalized linear model Generalized linear mixed model Glutathione S-transferases Knockdown resistance Magnesium Piperonyl butoxide Polymerase chain reaction Permethrin Total dissolved solids Tetramethylbenzidine Tap water Voltage-gated sodium channel World Health Organization Author contributions IKG conceived the idea for this study. JA, IKG, and GKA finalized the study design. IKG, GKA, SSA, AAL, DAB, SG, ABIA, JOJ, SOD, GAK, and AODY conducted the experiments. IKG, GKA, and JA complied and analysed the data. JA, SKD, and RP supervised the study. JA, DKA, RD, and SKD reviewed the manuscript. All authors read and approved the final manuscript. Funding No funding was received for this study. Data availability All data generated and/or analysed during this study are included in this published article and its supplementary information files and in the Figshare repository at 10.6084/m9.figshare.32500230. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Ibrahim K. Gyimah, Email: igyimah@noguchi.ug.edu.gh Jewelna Akorli, Email: jakorli@noguchi.ug.edu.gh. References Associated Data Supplementary Materials Supplementary Material 1 Supplementary Material 2 Supplementary Material 3 Supplementary Material 4 Supplementary Material 5 Supplementary Material 6 Data Availability Statement All data generated and/or analysed during this study are included in this published article and its supplementary information files and in the Figshare repository at 10.6084/m9.figshare.32500230.