Simulating the effects of clumped egg laying on mosquito population dynamics in relation to gene-drive interventions

Gene-drive based vector control methods are a rapidly developing tool in the fight against malaria. They utilise highly targeted insertions of genes to express specific traits, such as biases in offspring sex ratio or inhibited vector competence, which are preferentially inherited by copying themselves between chromosomes. Although theoretically selfsustaining, most gene-drive methods are sensitive to numerous aspects of local mosquito population dynamics. Often however, due to gaps in our knowledge, mathematical modelling of gene drive systems makes highly simplifying assumptions about key aspects of mosquito ecology. There is an urgent need to better understand fine scale population processes to improve predictions of the likely impact of gene-drive releases and refine development of target product profiles. The principal drivers of local mosquitoes dynamics can be traced to the larval stages, where density-dependent mortality in larval habitats is a key regulator of local adult density. In this study, we explored these drivers by simulating non-homogenous egg-laying over time and analysing its relationship with differing functional forms of density dependence. We developed a discrete-time stochastic model of mosquito population dynamics, which incorporated the random, temporal clumping of egg laying. The model was fitted to historical longitudinal mosquito trapping data using advanced particle MCMC methods. From this we were able to better quantify the relationship between clumping of egg laying and densitydependent regulation of larval populations, and thus refine estimates of the mosquito reproduction number, Rm – a key determinant of the predicted impact of gene-drive interventions. We find incorporation of clumping of egg-laying improves model fit, and reduces estimates of Rm, thus making establishment and spread of gene-drive constructs more likely than predicted by models assuming higher Rm value.