Ecology of gene drives: Organism fitness under larval competition determines gene drive performance

Gene drives are a promising technology to control disease vectors, such as mosquitoes. However, as these interventions aim to modify and/or control the population of disease vectors, a full understanding of how gene drives interact with the ecology of the target species is essential. In this study, we explore how larval competition determines gene drive performance. We first extend an agent-based model of the larval behaviour of Aedes aegypti mosquitoes to obtain detailed predictions for gene drive designs under realistic larval competition. Using Bayesian optimisation techniques, we explore the performance of different gene drive designs and find the Pareto frontier of optimal underdominant drives with different release thresholds. To understand what determines the structure of the Pareto frontier, we conduct both local and global sensitivity analyses to assess the role different parameters play in determining gene drive performance. We find that ecological parameters, which determine the behaviour of the organism, dominate over genetic parameters for much of the parameter space, although genetic parameters play an important role at the frontier. Finally, we develop stochastic and deterministic surrogate models to enable efficient comparison and simulation without imposing the computational costs of agent-based approaches.