A machine-learning approach to map landscape connectivity in <i>Aedes aegypti</i> with genetic and environmental data
Evlyn Pless, Norah P. Saarman, Jeffrey R. Powell, Adalgisa Caccone, Giuseppe Amatulli
Abstract
Significance Aedes mosquitoes are projected to continue expanding their ranges, which could expose millions more humans to the diseases they carry. The implementation of vector control methods ranging from traditional (e.g., insecticides) to cutting edge (e.g., genetic modification) could be improved with landscape connectivity maps and increased understanding of the factors that affect mosquito dispersal. Here we present an iterative random forest method for integrating genetic and environmental data to map landscape connectivity. We achieve a correlation of 0.83 between the model’s predicted genetic distance and actual genetic distance. We produce a genetic connectivity map for the southern tier of the United States and discuss important factors to consider in mosquito control, e.g., the release of genetically modified mosquitoes.