Models that simulate the effects of interventions on malaria vectors and transmission make assumptions about how mosquitoes move in the environment, such as isotropic behavior and no sex-related differences. These are applied to dispersal between households and villages and processes such as host-seeking and oviposition. Most models use mathematically convenient dispersal kernels based on these assumptions given the paucity of available data to better parameterize mosquito movement and the increase in complexity required. Consequently, there are few available methods to explore the effects of landscape and environmental factors on dispersal patterns. Understanding these patterns is key to optimizing control strategies, particularly for genetic control methods that involve releasing modified mosquitoes that compete with the natural mosquito population. We present a framework for modeling Anopheles gambiae s.l. movement mechanistically using available mark-release-recapture, biological, and ecological data and describe how it can be tailored for different locations and scenarios. We demonstrate its use for São Tomé and Príncipe and the Comoros, two candidate field sites for genetic control trials. Furthermore, we show the effects on these islands of elevation, land use, village/city proximity, and wind on predicted dispersal kernels and the implications for mosquito population dynamics. The resultant dispersal kernel is unique to the landscape of interest and is easily calibrated to field data measurements. Finally, we compare these results with genetic methods for inferring dispersal and connectivity between different mosquito populations on the islands and suggest future directions for the synthesis of these two data streams.
