Ecological diversity of Anopheles gambiae s.l. and insecticide resistance across refugee camps in Kenya

31 Malaria remains a major threat during humanitarian crises, necessitating targeted vector control 32 strategies informed by local vector dynamics. Between May and July 2023, we conducted 33 larval surveys in refugee settlements across Dadaab, Kakuma, and Kalobeyei (Kenya), 34 collecting Anopheles larvae. Genotyping of 728 specimen s revealed spatial variations in 35 species composition. Overall, Anopheles arabiensis was the dominant species (59%, n=42 6), 36 followed by Anopheles coluzzii (35%, n=25 2), and Anopheles rufipes (1%, n=7). In 37 Dadaab, An. arabiensis was overwhelmingly dominant (94%, n=352/37 4). In contrast, the 38 Kakuma/Kalobeyei complex was characterized by the co -occurrence of An. coluzzii (72%, 39 n=252/350) and An. arabiensis (22%, n=74/3 50), with An. rufipes exclusively found in 40 Kalobeyei (7%, n=6/89) (Figure 2B). Notably, no members of the Anopheles funestus group 41 or Anopheles stephensi were detected. However, approximately 5% of the larvae across the 42 sites could not be resolved molecularly. High frequencies of the L1014F kdr mutation, a 43 pyrethroid resistance marker, were detected in An. coluzzii (Kakuma: 50%; Kalobeyei: 63%) 44 and An. arabiensis (Kakuma: 10%; Kalobeyei : 30%) populations in Turkana County. 45 Interestingly, no kdr mutations were observed in the An. arabiensis population from Dadaab. 46 These findings highlight significant spatial diversity in vector species composition and 47 resistance profiles, with An. coluzzii emerging as a dominant, pyrethroid-resistant vector in the 48 Kakuma/Kalobeyei complex. The results underscore the urgent need for targeted interventions, 49 including resistance monitoring and alternative insecticide-based strategies, to mitigate malaria 50 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint transmission risks in fragile, humanitarian settings. Further studies are warranted to address 51 unidentified larval species and seasonal transmission dynamics. 52 53

Malaria; Anopheles coluzzii; insecticide resistance; refugee camps; vector 54 surveillance; Kenya 55 56

57 Conflict and the displacement of large populations are recognized as significant public health 58 risks. The rapid influx of refugees, coupled with limited sanitation infrastructure, often leads 59 to disease outbreaks. Infectious diseases are a major cause of morbidity and mortality in refugee 60 camps. Between 2009 and 2017, the United Nations High Commissioner for Refugees 61 (UNHCR) reported over 350 infectious disease outbreaks in refugee camps globally[1, 2], with 62 Kenya, Chad, and Thailand among the countries bearing the highest disease burdens. Vector -63 borne diseases, including yellow fever, dengue, and malaria, contribute substantially to this 64 burden during humanitarian crises[3-6]. 65 In Kenya, Kakuma and Dadaab refugee camps were established in response to humanitarian 66 crises stemming from conflict and famine in neighboring countries. Kakuma, located in 67 Turkana County in northwestern Kenya, was established in 1992 and is home to approximately 68 300,000 inhabitants, primarily refugees from South Sudan and Somalia [7, 8] . In 2016, the 69 UNHCR, in collaboration with the national and county governments of Turkana, established 70 the Kalobeyei Integrated Settlement to alleviate congestion in Kakuma and promote socio -71 economic integration with the host community. 72 Dadaab, established in 1991, is the largest refugee camp in Kenya, located in Garissa County 73 near the Kenyan-Somali border. With a population of approximately 400,000, predominantly 74 Somali refugees, Dadaab is administratively divided into three sub-camps: Ifo, Hagadera, and 75 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint Dagahaley[9]. These camps face significant challenges in managing public health risks, 76 particularly vector-borne diseases, due to overcrowding, limited resources, and the dynamic 77 nature of refugee populations. 78 Turkana County, Kenya, despite its semi -arid climate and low rainfall, experiences a 79 disproportionately high malaria burden [10-12], sustained by local, year -round transmission 80 that shows little correlation with rainfall[13]. This persistent threat is now compounded by the 81 recent discovery of two novel malaria vectors uniquely adapted to arid ecologies: the 82 invasive Anopheles stephensi [14]and the West African Anopheles coluzzii [15]. The larval 83 ecology of An. stephensi, in particular, is characterized by its specialization for breeding in a 84 wide range of human -made water containers including storage tanks, cisterns, and discarded 85 tires which are abundant in urban settings and essential for water security in drought -prone 86 areas.. This adaptability is further enhanced by its independence from rainfall and its ecological 87 flexibility to utilize natural sites like riverbeds. Together, these traits underpin its role as a major 88 and emerging urban malaria vector in Africa [16-18]. Similarly, An. coluzzii is strongly 89 associated with anthropogenic habitats in arid ecologies, with larvae typically found in 90 permanent or semi -permanent human -made aquatic habitats such as irrigated fields, urban 91 drainage, and stored water containers[19]. These adaptations make refugee camps like Kakuma 92 particularly vulnerable. Essential water harvesting and storage practices create abundant 93 breeding sites, a factor strongly linked to increased mosquito-borne disease transmission[20-94 22].The coexistence of An. coluzzii and An. stephensi with established vectors like An. 95 arabiensis suggests a complex and evolving transmission landscape, potentially exacerbating 96 malaria burden in Turkana and neighboring regions. 97 Understanding the bionomics and insecticide resistance profiles of these vectors is critical for 98 refining disease control strategies and especially within humanitarian contexts. In this study, 99 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint we examined the diversity and insecticide resistance profile of local malaria vector populations 100 of Kakuma and Dadaab refugee camps. 101 102

103 Larval surveys 104 Mosquito larvae were collected from randomly selected sites in Kakuma and Dadaab refugee 105 camps between May 2023 and July 2023 (Supplementary Table 1). Habitats were sampled once 106 during the study period. Larvae at the L1-L3 developmental stages were sampled from a variety 107 of habitats, including water containers, stagnant water, and roadside pools, using a standard 108 350 mL dipper. At each site, at least three dips were taken to quantify larval density. Anopheles 109 larvae were separated and individually preserved in 1.5 mL microcentrifuge tubes containing 110 95% ethanol for transport to the KEMRI Wellcome Trust Research Programme in Kilifi, 111 Kenya. 112 Nucleic acid extraction and molecular analysis for species identification 113 Field-collected larvae were rinsed with nuclease-free water, and genomic DNA extracted using 114 the Chelex method. Briefly, whole larvae were transferred into individual 1.5 mL 115 microcentrifuge tubes containing 50 µL of 20% Chelex resin (Bio-Rad, USA) and 116 homogenized using polypropylene pestles. The lysate was incubated at 100 °C while shaking 117 at 650 rpm on a ThermoMixer (Eppendorf, Hamburg, Germany). The solution was centrifuged 118 at 10,000 × g for 2 minutes, and the supernatant transferred to a new 1.5 mL microcentrifuge 119 tube. This process was repeated twice, and the extracted DNA stored at −80 °C until further 120 analysis. Anopheles gambiae s.l. sibling species were identified using a previously described 121 PCR method targeting the intergenic spacer (IGS) region of the ribosomal DNA [23]. Samples 122 that did not amplify for An. gambiae s.l. were screened for members of the An. funestus 123 complex using established protocols[24] and An. stephensi using established protocols [14]. 124 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint Nucleic acid extraction and molecular analysis for species identification 125 Field-collected larvae were rinsed with nuclease-free water, and genomic DNA was extracted 126 using the Chelex method. Briefly, whole larvae were transferred into individual 1.5 mL 127 microcentrifuge tubes containing 50 µL of 20% Chelex resin (Bio -Rad, USA) and 128 homogenized using polypropylene pestles. The lysate was incubated at 100 °C while shaking 129 at 650 rpm on a ThermoMixer (Eppendorf, Hamburg, Germany). The solution was centrifuged 130 at 10,000 × *g* for 2 minutes, and the supernatant was transferred to a new 1.5 mL 131 microcentrifuge tube. This process was repeated twice, and the extracted DNA was stored at 132 −80 °C until further analysis. 133 Species identification was performed using a sequential molecular workflow. First, Anopheles 134 gambiae s.l. sibling species were identified using a PCR method targeting the intergenic spacer 135 (IGS) region of the ribosomal DNA [23]. Samples identified as An. gambiae s.s. were further 136 characterized to distinguish between the M (An. coluzzii) and S (An. gambiae) forms via a PCR 137 assay targeting the SINE200 insertion/deletion polymorphism on the X chromosome [25]. The 138 primers used were: Forward: 5′ -GTGTGCACCTCGACGTAC-3′ and Reverse: 5′ -139 CGGAGTGACCAGGACACC-3′. PCR products were resolved on a 2% agarose gel; samples 140 showing a band at ~315 bp were classified as An. coluzzii, those with a band at ~250 bp as An. 141 gambiae, and those with both bands as hybrid forms. Samples that did not amplify for An. 142 gambiae s.l. were subsequently screened for members of the An. funestus complex [24] and 143 for An. stephensi [14]. This sequential identification workflow was chosen based on historical 144 data and previous studies indicating the dominance of the An. gambiae complex over the An. 145 funestus group in these arid ecological zones [26]. 146 147 Molecular Identification of PCR-Unresolved Mosquito Specimens 148 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint For mosquitoes unresolved by initial PCR, the ITS2 region was amplified using primers from 149 Beebe & Saul (1995)[2]. The 10 µL PCR reaction contained 5 µL of 2X GoTaq® Master Mix 150 (Promega, USA), 0.5 µL of each primer, 1.5 µL of DNA, and 2.5 µL nuclease-free water. 151 Cycling conditions were: 95°C for 5 min; 40 cycles of 95°C for 15 sec, 52°C for 20 sec, and 152 72°C for 1 min; with a final 72°C for 10 min. Amplicons were purified with the QIAquick PCR 153 Purification Kit (Qiagen, Germany). Sanger sequencing used BigDye Terminator v3.1 (Applied 154 Biosystems, UK) on an ABI 3730xl sequencer. Chromatograms were edited in CLC Main 155 Workbench 24. Consensus sequences were identified via BLASTn against the NCBI nt 156 database under default parameters[27]. 157 158 Insecticide resistance profiling 159 The presence of point mutations at position 1014 of the voltage -gated sodium channel (vgsc) 160 gene (knock down resistance (kdr) mutations) were investigated using a TaqMan probe-based 161 quantitative real-time PCR (qPCR) assay. The assay utilized one set of primers: Forward: 5′ -162 CAT TTT TCT TGG CCA CTG TAG TGA T-3′, Reverse: 5′-CGA TCT TGG TCC ATG TTA 163 ATT TGC A-3′ and three probes for detection of: wild -type allele: 5′ -CTT ACG ACT AAA 164 TTT C-3′ (labeled with HEX fluorophore), Vgsc -L1014F mutation: 5′-ACG ACA AAA TTT 165 C-3′ (labeled with FAM fluorophore) and Vgsc-L1014S mutation: 5′-ACG ACT GAA TTT C-166 3′ (labeled with FAM fluorophore). The qPCR cycling conditions consisted of an initial 167 denaturation at 95 °C for 10 minutes, followed by 40 cycles of denaturation at 95 °C for 10 168 seconds and annealing and extension at 65 °C for 45 seconds. A sample is 169 considered positive for a specific allele (wild-type, L1014F, or L1014S) if its associated probe 170 generates a fluorescence signal above the threshold during qPCR. 171 172

173 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint Vector abundance, distribution and taxonomic assignment 174 Larval surveillance conducted in May and July 2023 identified productive Anopheles breeding 175 habitats in both the Dadaab (Garissa County) and Kakuma/Kalobeyei (Turkana County) 176 refugee camps (Figure 1A). The survey of 13 habitats revealed an overall mean density of 6.6 177 larvae/dip, though this varied significantly between camp environments (Table 1). 178 In Dadaab, larval production was confined to anthropogenic habitats. Vehicle washing 179 bays and water-filled vehicle ruts were the primary larval sources, yielding a sub -total of 457 180 larvae. While vehicle ruts were the most prolific habitat type in this camp (mean density: 8.5 181 larvae/dip), the overall mean density for Dadaab was 5.1 larvae/dip (Table 1). 182 In contrast, breeding sites in Kakuma and Kalobeyei were more varied and productive, 183 comprising natural and peri -domestic habitats (Figure 1B&C). These sites yielded a higher 184 overall mean density of 10.0 larvae/dip. As detailed in Table 1, the most productive habitat 185 categories were streams/riverbeds (mean density: 32.5 larvae/dip) and roadside ponds (mean 186 density: 24.6 larvae/dip). The single most productive habitat identified in the entire study was 187 a roadside pond in Kakuma, which reached a density of 57.5 larvae/dip. The highly variable 188 productivity of roadside drainages, which ranged from zero to 16.8 larvae/dip, underscores the 189 heterogeneous distribution of breeding sites within the camp environment. 190 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint 191 Figure1 : Malaria vector larval sampling in Kakuma and Dadaab refugee complex 192 A)Map of sites sampled for anopheline larvae; B and C) larval breeding sites 193 194 From 877 larvae collected , DNA was extracted successfully from 728 and subjected to 195 molecular genotyping for An. gambiae, An. funestus species complexes and An. stephensi and 196 ITS-2 amplicon sequencing . Notably, no specimens of the Anopheles funestus group 197 or Anopheles stephensi were identified in any of the camps . Overall, Anopheles arabiensis 198 dorminated larval collection in refugee camps(59%) followed by An. coluzzii (35%), An rufipes 199 (1%) and lastly a single specimen of An. gambiae s.s. identified. Approximately 5% of 200 mosquitoes collected could not be identified by combination of PCR based methods nor ITS-2 201 sequencing using the sanger sequencing approach (Figure 2A, Table S1 and Table S2). 202 The vectors identified were spatially distributed as follows. In Dadaab, An. arabiensis was the 203 predominant species, accounting for 9 4% (n = 3 52) of the collections, with only a single 204 specimen of An. gambiae s.s. identified. In Kakuma, An. coluzzii. dominated (84%, n = 222), 205 while An. arabiensis represented 10% (n = 2 7). In Kalobeyei, An. arabiensis (54%, n=47) 206 and An. coluzzii (34%, n=30) and An. rufipes (7%, n=6) were the primary species (Figure 2B). 207 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint In addition, samples from western Kenya (Luyeshe and Maseno) were included for 208 comparative purposes and only identified An. gambiae s.s. (Figure S1). 209 210 Figure 2. Species composition of Anopheles mosquitoes in Kakuma, Kalobeyei, and 211 Dadaab refugee camps, Kenya . (A) Overall relative abundance of Anopheles 212 species identified by PCR and ITS2 sequencing. (B) Spatial distribution and abundance of the 213 predominant species, An. coluzzii and An. arabiensis, across individual sampling sites. 214 215 Frequency of kdr mutations in vgsc of An. gambiae s.l. 216 The kdr-L1014F mutation was detected at high frequencies in An. coluzzii from Kakuma and 217 Kalobeyei, with allelic frequencies of 5 0% and 63 %, respectively. In contrast, An. 218 arabiensis exhibited lower frequencies of the L1014F mutation (10% and 30% in Kakuma and 219 Kalobeyei, respectively). Neither the L1014S nor L1014F mutation was observed in specimens 220 from Dadaab. High levels of insecticide resistance mutations were thus limited to An. coluzzii. 221 in Turkana County and absent in Garissa (Table 2). 222 223 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint

224 Malaria outbreaks in Turkana County have increased in recent years, with year -round P . 225 falciparum transmission that is not strongly correlated with rainfall patterns [28]. 226 Additionally, P . vivax infections have been reported among semi -nomadic populations in 227 Turkana and northeastern Kenya[29] who have been implicated in facilitating parasite spread 228 through their migration routes. Shifts in mosquito vector populations are known to influence 229 malaria transmission [30, 31]. Understanding bionomics of vector populations at the local scale 230 is a critical first step for elucidating transmission dynamics and informing targeted intervention. 231 This study investigated the larval ecology of malaria vectors in Kakuma and Dadaab refugee 232 camps, revealing significant changes in vector composition. In Kakuma this represents a 233 significant shift from the previously documented vector population from 2005 -2006, which 234 was reported to be homogeneously An. arabiensis [26]. The population has now changed to a 235 more heterogeneous population comprising An. coluzzii, An. arabiensis and An rufipes two 236 decades later. Notably, An. rufipes was identified through the ITS2 sequencing approach after 237 being missed by standard PCR, underscoring the value of complementary molecular methods 238 for accurate species resolution. The role of this vector in the Kakuma -Kalobeyei complex 239 cannot be overlooked, as it may contribute to the local transmission dynamics. While malaria 240 vector populations shifts have been documented in various settings, they have tended towards 241 a decline in An. gambiae s.s and an increase in An. funestus as the dominant species[32]. The 242 absence of An. funestus and An. stephensi in our larval collections, despite their known 243 presence in other parts of Kenya, suggests that the specific breeding habitats in these camps—244 dominated by temporary sunlit pools and vehicle ruts may not be suitable for these species, 245 which prefer more permanent, vegetated, or container -based habitats respectively . These 246 ecological shifts are largely attributed to widespread insecticide use[32]. Despite year-round 247 malaria transmission, vector surveillance In Kakuma has been limited , leaving the primary 248 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint malaria vectors poorly characterized. Concurrently, the camp's human population has surged 249 over the past two decades, prompting establishment of the Kalobeyei settlement. 250 These anthropogenic changes, along with environmental and ecological factors may have 251 created conditions conducive for proliferation of An. coluzzii [33]. It is theori zed that the 252 speciation of An. gambiae and An. coluzzii is driven not only by genetics but ecological factors 253 including quality of aquatic habitats[19]. The later emergence of An. coluzzii is associated with 254 the diversification of aquatic habitats including those of marginal quality, for which it appears 255 predisposed[19]. This is to some extent reflected in the diversity and unusual habitats within 256 which Anophelines were found breeding in Kakuma. With regards to infection, susceptibility 257 to Plasmodium falciparum of all three vector species varies depending on local environment, 258 genetic factors and parasite strain. However, high P . falciparum infection rates have been 259 reported in An. coluzzii and in one study P . vivax infection was documented [34]. In Kakuma, 260 the current vector composition represents a notable change from the previously documented 261 population from 2005-2006, which was reported to be homogeneously An. arabiensis [26]. The 262 population now comprises a more heterogeneous mixture of An. coluzzii , An. arabiensis , 263 and An. rufipes. Given the lack of continuous longitudinal data, this change could represent 264 either a local population shift or the recent introduction and subsequent establishment of An. 265 coluzzii in the region . The presence and establishment or expansion of An. coluzzii in the 266 region appears to be a relatively recent phenomenon, as it was not detected during vector 267 surveillance in Kakuma fifteen years ago [26]. Therefore, its role in past and current malaria 268 outbreaks in Turkana County is still poorly understood warranting further investigation, 269 particularly given conflicting evidence on its vectorial capacity and susceptibility 270 to Plasmodium infections[35, 36]. 271 In contrast, An. arabiensis remained the dominant vector in Dadaab. Unlike Kakuma, Dadaab 272 experiences little to no malaria transmission. The underlying mechanisms for this low 273 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint transmission, despite the presence of a competent vector, are not fully understood. Potential 274 contributing factors could include the zoophilic tendencies of local An. arabiensis populations, 275 the influence of endosymbiotic microorganisms, or other genetic, environmental, and 276 anthropogenic factors that may reduce transmission efficiency [37, 38]. However, the specific 277 drivers in this context remain unclear, underscoring the need for dedicated studies to elucidate 278 the malaria transmission dynamics and the relative contribution of different factors in Dadaab 279 Overall, the identity of approximately 5% of the samples could not be resolved using standard 280 An. gambiae s.l. and An. funestus s.l. species complex, An stephensi species specific PCR or 281 ITS2 amplicon sequencing identification protocols. This includes definitively ruling out An. 282 stephensi, for which we employed a s pecific PCR assay [14], confirming it was not present 283 among these unresolved specimens. The fact that An. rufipes was only detected through 284 sequencing highlights a key limitation of standard PCR assays and highlights the critical need 285 to employ complementary sequencing and genomic tools to fully characterize anopheline 286 diversity. This approach is essential to uncover potentially overlooked secondary vectors whose 287 role in transmission may currently be underestimated. 288 Regarding insecticide resistance, high frequencies of the kdr-L1014F mutation were observed 289 in An. coluzzii from Kakuma and Kalobeyei suggesting a move towards fixation , with most 290 mosquitoes exhibiting either homozygous or heterozygous genotypes. Lower frequencies of 291 the mutation were detected in An. arabiensis from the same sites, while no kdr mutations were 292 found in specimens from Dadaab. These findings suggest selecti on pressure favoring 293 insecticide resistance genotypes in Turkana County, potentially driven by local insecticide use 294 or the introduction of resistant mosquitoes from neighboring regions, such as eastern Uganda, 295 where high levels of resistance have been documented[39]. However, this study focused solely 296 on genotypic resistance; phenotypic assays are needed to confirm the susceptibility of these 297 populations to insecticides used in vector control programs. 298 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint Long-lasting insecticidal nets (LLINs) and indoor residual spraying (IRS) remain critical for 299 malaria control in refugee camps, but their efficacy depends on mosquito susceptibility to 300 insecticides. The emergence of pyrethroid resistance in Turkana County , evidenced by 301 high kdr-L1014F frequencies, demands integrated vector management strategies 302 and continuous monitoring of kdr allele frequencies to mitigate this threat. 303 While our study provides key insights into larval ecology across three refugee settlements, we 304 acknowledge limitations inherent to larval-based surveillance. Larval collections are subject to 305 sampling bias, as cryptic breeding sites, particularly those of Anopheles funestus and secondary 306 vectors, often evade detection despite confirmed adult presence. Consequently, our data reflect 307 larval diversity within accessible habitats rather than exhaustive vector composition. 308 Nevertheless, in Turkana (Kakuma/Kalobeyei) and Dadaab, where anthropogenic sites (water 309 storage containers, drainage ditches) sustain dominant An. coluzzii and An. 310 arabiensis populations, larval source management (LSM) emerges as a high -priority 311 intervention. Targeted LSM (e.g., container covering, habitat drainage, site -specific 312 larviciding) could effectively suppress these container -breeding gambiae-complex vectors. In 313 contrast, settings with confirmed funestus-group dominance would require alternative 314 approaches due to their association with cryptic, vegetated aquatic habitats. Thus, our findings 315 strongly support LSM in semi-arid humanitarian contexts where gambiae-complex vectors 316 dominate identifiable artificial habitats, while underscoring the need for complementary adult 317 surveillance to resolve cryptic vector dynamics. 318 319

320 This study documents significant ecological shifts in malaria vectors across Kenya's arid-zone 321 refugee settlements, revealing the emergence of Anopheles coluzzii alongside An. arabiensis in 322 Turkana County , linked to anthropogenic habitat modifications that enable year -round 323 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint transmission—while Dadaab exhibits persistent An. arabiensis dominance without 324 proportional malaria burden, suggesting unexplained local refractoriness. Critically, high kdr-325 L1014F frequencies in Turkana's An. coluzzii (upto 63%) pose an immediate threat to 326 pyrethroid-based interventions (LLINs/IRS), necessitating urgent resistance management 327 through next-generation tools and phenotypic validation. Furthermore, the unresolved identity 328 of 5% of specimens underscores the need for enhanced surveillance using whole -genome 329 sequencing or 330 Matrix-Assisted Laser Desorption/Ionization Time -of-Flight (MALDI-TOF). These findings 331 collectively highlight the urgency for habitat-focused vector control in semi-arid humanitarian 332 contexts: where gambiae-complex vectors dominate identifiable artificial breeding sites (water 333 storage containers, drainage ditches), targeted larval source management (container covering, 334 site-specific larviciding) offers a critical strategy to disrupt transmission at its source, 335 complementing insecticide resistance countermeasures to protect crisis-affected populations. 336 337

338 We thank the technical and field staff: Festus Yaa, Gabriel Nzai, and Julius Tineja, who helped 339 with mosquito sample collection in the field. 340 341 Author contributions 342 M.K.R. conceptualisation, methodology, formal analysis, funding acquisition, writing —343 original draft. M.M. conceptuali sation, methodology, reviewing and editing. C.W. 344 methodology, reviewing and editing. L.I.O. funding acquisition and reviewing and editing. J.M. 345 reviewing and editing. R.W.S. reviewing and editing. B.B., K.G., A.D., M.T. and A.O. sample 346 collection and processing, laboratory analysis and reviewing and editing. All authors read and 347 approved the manuscript. 348 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint 349 Funding 350 This work is supported by The Royal Society FLAIR fellowship grant: FLR \R1\190497, 351 FCG\R1\211043 and KEMRI IRG grant: KEMRI \IRG\NN02 (awarded to M.K.R.). and 352 molecular reagent support from a pathogen genomics sub -Award (to L.I.O) from Africa CDC 353 PGI and ASLM. RWS is supported by the Wellcome Trust Principal Fellowship (#212176) . 354 All authors are grateful for the support of the Wellcome Trust to the Kenya Major Overseas 355 Programme (#203077). 356 357 Ethics approval and consent to participate 358 The study was approved by the KEMRI Scientific and Ethics Review Unit (SERU) with the 359 protocol number: 337 KEMRI/SERU/CGMR-C/024/3148. 360 361 Consent for publication 362 This manuscript is published with the permission of the Director-General of the Kenya Medical 363 Research Institute.The funding bodies had no role in the design, data collection, and drafting 364 of the manuscript. 365 366 Competing interests 367 All authors have declared there are no competing interests. 368 369 Availability of data and materials 370 Most of the dataset used for analysis is available in the manuscript. We withheld the geo-data 371 that may predispose individual homesteads to a high risk of identifiability. However, they are 372 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint under the custodianship of the KEMRI Wellcome Trust Data Governance Committee and are 373 accessible upon request addressed to that committee. 374 375 Tables 376 Table 1. Summary of Anopheles larval habitat type in refugee camps in Kenya. 377 378 Camp & Habitat No. of sites Total Dips Total Larvae Mean Larvae/Dip (Range) Dadaab Vehicle Washing Bay 3 50 117 2.3 (0.0 - 3.8) Vehicle Ruts 2 40 340 8.5 (2.0 - 15.0) Subtotal 5 90 457 5.1 Kakuma/Kalobeyei Stream/River 2 4 130 32.5 (23.5 - 41.5) Roadside Pond 2 5 123 24.6 (2.7 - 57.5) Roadside Drainage 4 33 167 5.1 (0.0 - 16.8) Subtotal 8 42 420 10.0 Overall Total 13 132 877 6.6 379 Table 2. kdr-West (L1014F) genotype distribution and allelic frequencies in Anopheles 380 populations across study sites 381 Site kdr-West(L1014F) Species Genotype count Allelic frequency # RR RS SS R S Dadaab An. arabiensis 322 0 0 322 0 1 kakuma An. coluzzii 58 3 52 3 0.5 0.5 Kakuma An. arabiensis 26 5 4 17 0.27 0.73 Kalobeyei An. coluzzii 27 9 16 2 0.63 0.37 Kalobeyei An. arabiensis 40 0 8 32 0.1 0.9 382

383 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint 1. Chiara Altare VK, Moise Ngwa, Amelia Goldsmith, Heiko Hering, Ann Burton, 384 Paul Spiegel: Infectious disease epidemics in refugee camps: a retrospective 385 analysis of UNHCR data (2009-2017). Journal of Global Health Reports 2019, 3. 386 2. Beebe NW, Saul A: Discrimination of all members of the Anopheles 387 punctulatus complex by polymerase chain reaction--restriction fragment 388 length polymorphism analysis. Am J Trop Med Hyg 1995, 53:478-481. 389 3. Arcos Gonzalez P , Cabria Fernandez J, Gan RK, Fernandez Camporro A, Cernuda 390 Martinez JA: The epidemiological profile of incidence and mortality from 391 epidemics in complex humanitarian emergencies from 1990 to 2022 - A 392 scoping review. Trop Med Int Health 2024, 29:343-353. 393 4. Burki T: Infectious diseases in Malian and Syrian conflicts. Lancet Infect Dis 394 2013, 13:296-297. 395 5. Orach CG: Morbidity and mortality amongst southern Sudanese in Koboko 396 refugee camps, Arua District, Uganda. East Afr Med J 1999, 76:195-199. 397 6. Xu JW, Liu H, Yaw B, Nbwi HS: The health beliefs, dengue knowledge and 398 control behaviors among internally displaced persons versus local residents 399 in Kachin Special Region II, Myanmar. PLoS Negl Trop Dis 2020, 14:e0008321. 400 7. UNHCR: THE KALOBEYEI INTEGRATED SOCIO-ECONOMIC DEVELOPMENT 401 PROGRAMME :KISEDP . (UNHCR ed.2018. 402 8. Kakuma Refugee Camp [https://www.unhcr.org/ke/about-us/where-we-403 work/kakuma-refugee-camp] 404 9. Kimani EW, Vulule JM, Kuria IW, Mugisha F: Use of insecticide-treated clothes 405 for personal protection against malaria: a community trial. Malar J 2006, 406 5:63. 407 10. Alegana VA, Macharia PM, Muchiri S, Mumo E, Oyugi E, Kamau A, Chacky F , 408 Thawer S, Molteni F , Rutazanna D, et al: Plasmodium falciparum parasite 409 prevalence in East Africa: Updating data for malaria stratification. PLOS Glob 410 Public Health 2021, 1:e0000014. 411 11. MOH K: Kenya Malaria Strategy 2023-2027. pp. 119. Nairobi, Kenya: Ministry of 412 Health; 2024:119. 413 12. Noor AM, Gething PW, Alegana VA, Patil AP , Hay SI, Muchiri E, Juma E, Snow RW: 414 The risks of malaria infection in Kenya in 2009. BMC Infect Dis 2009, 9:180. 415 13. Markwalter CF , Menya D, Wesolowski A, Esimit D, Lokoel G, Kipkoech J, 416 Freedman E, Sumner KM, Abel L, Ambani G, et al: Plasmodium falciparum 417 importation does not sustain malaria transmission in a semi-arid region of 418 Kenya. PLOS Glob Public Health 2022, 2:e0000807. 419 14. Ochomo EO, Milanoi S, Abong'o B, Onyango B, Muchoki M, Omoke D, Olanga E, 420 Njoroge L, Juma EO, Otieno JD, et al: Detection of Anopheles stephensi 421 Mosquitoes by Molecular Surveillance, Kenya. Emerg Infect Dis 2023, 422 29:2498-2508. 423 15. Kamau L, Bennett KL, Ochomo E, Herren J, Agumba S, Otieno S, Omoke D, 424 Matoke-Muhia D, Mburu D, Mwangangi J, et al: The Anopheles coluzzii range 425 extends into Kenya: detection, insecticide resistance profiles and 426 population genetic structure in relation to conspecific populations in West 427 and Central Africa. Malar J 2024, 23:122. 428 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint 16. Hawaria D, Kibret S, Zhong D, Lee MC, Lelisa K, Bekele B, Birhanu M, Mengesha 429 M, Solomon H, Yewhalaw D, Yan G: First report of Anopheles stephensi in 430 Southern Ethiopia. Res Sq 2023. 431 17. Tadesse FG, Ashine T, Teka H, Esayas E, Messenger LA, Chali W, Meerstein-432 Kessel L, Walker T, Wolde Behaksra S, Lanke K, et al: Anopheles stephensi 433 Mosquitoes as Vectors of Plasmodium vivax and falciparum, Horn of Africa, 434 2019. Emerg Infect Dis 2021, 27:603-607. 435 18. Yared S, Gebresilassie A, Aklilu E, Abdulahi E, Kirstein OD, Gonzalez-Olvera G, 436 Che-Mendoza A, Bibiano-Marin W, Waymire E, Lines J, et al: Building the vector 437 in: construction practices and the invasion and persistence of Anopheles 438 stephensi in Jigjiga, Ethiopia. Lancet Planet Health 2023, 7:e999-e1005. 439 19. Costantini C, Ayala D, Guelbeogo WM, Pombi M, Some CY , Bassole IH, Ose K, 440 Fotsing JM, Sagnon N, Fontenille D, et al: Living at the edge: biogeographic 441 patterns of habitat segregation conform to speciation by niche expansion in 442 Anopheles gambiae. BMC Ecol 2009, 9:16. 443 20. Biswas S, Rajkonwar J, Nirmolia T, Jena SR, Sarkar U, Bhattacharyya DR, 444 Borkakoty B, Pandey A, Subbarao SK, Majumder T, et al: First Report of Rubber 445 Collection Bowls & Plastic and Bamboo Water Containers as the Major 446 Breeding Source of Ae. albopictus with the Indigenous Transmission of 447 Dengue and Chikungunya in Rural Forested Malaria-Endemic Villages of 448 Dhalai District, Tripura, India: The Importance of Molecular Identification. 449 Biomedicines 2023, 11. 450 21. Mwangangi JM, Muturi EJ, Mbogo CM: Seasonal mosquito larval abundance 451 and composition in Kibwezi, lower eastern Kenya. J Vector Borne Dis 2009, 452 46:65-71. 453 22. Kassahun Waktola D: Malaria and pond-based rainwater harvesting linkages 454 in the fringes of central highland Ethiopia. Rural Remote Health 2008, 8:956. 455 23. Scott JA, Brogdon WG, Collins FH: Identification of single specimens of the 456 Anopheles gambiae complex by the polymerase chain reaction. Am J Trop 457 Med Hyg 1993, 49:520-529. 458 24. Koekemoer LL, Kamau L, Hunt RH, Coetzee M: A cocktail polymerase chain 459 reaction assay to identify members of the Anopheles funestus (Diptera: 460 Culicidae) group. Am J Trop Med Hyg 2002, 66:804-811. 461 25. Santolamazza F , Mancini E, Simard F , Qi Y , Tu Z, della Torre A: Insertion 462 polymorphisms of SINE200 retrotransposons within speciation islands of 463 Anopheles gambiae molecular forms. Malar J 2008, 7:163. 464 26. Bayoh MN, Akhwale W, Ombok M, Sang D, Engoki SC, Koros D, Walker ED, 465 Williams HA, Burke H, Armstrong GL, et al: Malaria in Kakuma refugee camp, 466 Turkana, Kenya: facilitation of Anopheles arabiensis vector populations by 467 installed water distribution and catchment systems. Malar J 2011, 10:149. 468 27. Altschul SF , Gish W, Miller W, Myers EW, Lipman DJ: Basic local alignment 469 search tool. J Mol Biol 1990, 215:403-410. 470 28. Meredith HR, Wesolowski A, Okoth D, Maraga L, Ambani G, Chepkwony T, Abel L, 471 Kipkoech J, Lokoel G, Esimit D, et al: Characterizing mobility patterns and 472 malaria risk factors in semi-nomadic populations of Northern Kenya. PLOS 473 Glob Public Health 2024, 4:e0002750. 474 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint 29. O'Meara WP , Maraga L, Meredith H, Esimit D, Lokoel G, Chepkwony T, Kipkoech J, 475 Ambani G, Menya D, Freedman E, et al: Plasmodium vivax Prevalence in 476 Semiarid Region of Northern Kenya, 2019. Emerg Infect Dis 2023, 29:2385-477 2387. 478 30. Cross DE, Thomas C, McKeown N, Siaziyu V , Healey A, Willis T, Singini D, Liywalii 479 F , Silumesii A, Sakala J, et al: Geographically extensive larval surveys reveal 480 an unexpected scarcity of primary vector mosquitoes in a region of 481 persistent malaria transmission in western Zambia. Parasit Vectors 2021, 482 14:91. 483 31. Kreppel KS, Viana M, Main BJ, Johnson PCD, Govella NJ, Lee Y , Maliti D, Meza FC, 484 Lanzaro GC, Ferguson HM: Emergence of behavioural avoidance strategies of 485 malaria vectors in areas of high LLIN coverage in Tanzania. Sci Rep 2020, 486 10:14527. 487 32. Msugupakulya BJ, Urio NH, Jumanne M, Ngowo HS, Selvaraj P , Okumu FO, 488 Wilson AL: Changes in contributions of digerent Anopheles vector species to 489 malaria transmission in east and southern Africa from 2000 to 2022. Parasit 490 Vectors 2023, 16:408. 491 33. Hinne IA, Attah SK, Mensah BA, Forson AO, Afrane YA: Larval habitat diversity 492 and Anopheles mosquito species distribution in digerent ecological zones in 493 Ghana. Parasit Vectors 2021, 14:193. 494 34. Ibrahim SS, Mukhtar MM, Irving H, Labbo R, Kusimo MO, Mahamadou I, Wondji 495 CS: High Plasmodium infection and multiple insecticide resistance in a 496 major malaria vector Anopheles coluzzii from Sahel of Niger Republic. Malar 497 J 2019, 18:181. 498 35. Boissiere A, Gimonneau G, Tchioho MT, Abate L, Bayibeki A, Awono-Ambene PH, 499 Nsango SE, Morlais I: Application of a qPCR assay in the investigation of 500 susceptibility to malaria infection of the M and S molecular forms of An. 501 gambiae s.s. in Cameroon. PLoS One 2013, 8:e54820. 502 36. Fryxell RT, Nieman CC, Fofana A, Lee Y , Traore SF , Cornel AJ, Luckhart S, Lanzaro 503 GC: Digerential Plasmodium falciparum infection of Anopheles gambiae s.s. 504 molecular and chromosomal forms in Mali. Malar J 2012, 11:133. 505 37. Gnémé A, Guelbéogo WM, Riehle MM, Sanou A, Traoré A, Zongo S, Eiglmeier K, 506 Kabré GB, Sagnon NF , Vernick KD: Equivalent susceptibility of Anopheles 507 gambiae M and S molecular forms and Anopheles arabiensis to Plasmodium 508 falciparum infection in Burkina Faso. Malaria Journal 2013, 12:204. 509 38. Ahouandjinou MJ, Sovi A, Sidick A, Sewade W, Koukpo CZ, Chitou S, Towakinou 510 L, Adjottin B, Hougbe S, Tokponnon F , et al: First report of natural infection of 511 Anopheles gambiae s.s. and Anopheles coluzzii by Wolbachia and 512 Microsporidia in Benin: a cross-sectional study. Malar J 2024, 23:72. 513 39. Tchouakui M, Oruni A, Assatse T, Manyaka CR, Tchoupo M, Kayondo J, Wondji CS: 514 Fitness cost of target-site and metabolic resistance to pyrethroids drives 515 restoration of susceptibility in a highly resistant Anopheles gambiae 516 population from Uganda. PLoS One 2022, 17:e0271347. 517 518 .CC-BY 4.0 International licenseperpetuity. It is made available under a preprint (which was not certified by peer review) is the author/funder, who has granted bioRxiv a license to display the preprint in The copyright holder for thisthis version posted October 15, 2025. ; https://doi.org/10.1101/2025.10.13.682237doi: bioRxiv preprint