Abstract
BACKGROUND: Malaria transmission, characterized by spatial and temporal heterogeneity and complex vector behaviours, persists in Kenya's highlands despite widespread use of long-lasting insecticidal nets (LLINs). The role of human activity in exposure risk remains underexplored. Identifying vulnerable times and locations is crucial for designing and optimizing targeted control strategies that address the intricate interplay between human activity and local vector behaviour that results in transmission. This study examined human-mosquito interactions in three different ecological settings in Nandi highlands in western Kenya. METHODS: Malaria vector biting rates were monitored both indoors and outdoors from 18:00 to 06:00 over five consecutive nights in ten houses per village in three different ecological settings namely site close to the forest (Kipsamoite), neutral site neither close to forest nor swamp (Kebulonik), site close to the swamp and with past high malaria prevalence (Kapsisywa) using human landing catches (HLC) during the long (May 2018) and short (October 2018) rainy seasons. Concurrently, hourly human behaviour observations (HBOs) were conducted to assess indoor versus outdoor presence, sleeping patterns and LLINs use. All Anopheles mosquitoes were first identified morphologically using standard anopheline keys and subsequently confirmed to species level through molecular sequencing of the internal transcribed spacer 2 (ITS2) region and cytochrome c oxidase subunit 1 (CO1) gene. RESULTS: High Anopheles species diversity was observed, with site-specific dominance: Anopheles arabiensis in Kipsamoite, Anopheles christyi in Kebulonik, and the novel Anopheles spp. 14 BSL-2014 in multiple sites. The majority of specimens were collected indoors in Kipsamoite (67%) and Kebulonik (52.9%), while most specimen were collected outdoors in Kapsisywa (58.3%) were outdoors. Mosquito exposure peaked outdoors in the early evening (1800-2100 h) and indoors during the first half of the night (1900-0100 h), coinciding with periods when people were awake or transitioning to or from sleep, with low LLIN use. Human behaviour-adjusted exposure was highest outdoors in the early evening (1800-2100 h) and indoors during the first half of the night (1900-0100 h). Overall, most exposure occurred indoors for unprotected sleepers and individuals awake (53-55%), followed by outdoor exposure in the early evening and late morning (16-44%). LLINs prevented 24.5% (long rains) and 44.9% (short rains) of bites in Kipsamoite, 24.6 to 37% in Kebulonik, and 35.8% in Kapsisywa (short rains). CONCLUSION: This study demonstrates that human exposure to malaria vectors is shaped by the interplay between temporal and spatial human and vector behaviours, with the highest biting rates indoors for unprotected sleepers and awake individuals, and outdoor exposure peaking in the early evening and late morning. It also reveals diverse, behaviourally adaptable vector populations, including cryptic species, sustaining indoor and outdoor transmission. While LLINs use provide partial protection, significant gaps in protection remained during periods and in spaces where nets are not effective, highlighting persistent residual transmission and the need for vector characterization, behaviour-informed interventions (e.g., spatial repellents and larviciding), community engagement, and strengthened entomological surveillance to guide effective malaria control.