Inferring intra-vector arboviral dynamics challenges a classical assumption used in most vector-borne disease models
Arboviruses are pathogens transmitted between vertebrate hosts by the bite of arthropod vectors, such as mosquitoes. Most of these viruses cause zoonotic diseases, which account for a significant proportion of emerging diseases worldwide, making them a threat to public health. Arbovirus transmission is a dynamic, multi-scale process, ranging from intra-vector viral dynamics to epidemic dynamics. Processes involved are influenced by numerous biotic and abiotic factors and are key elements in pathogen transmission. However, the dynamics of intra-vector viral infection (IVD) are rarely modelled explicitly. Rather, an average duration of the extrinsic incubation period is considered, assuming an exponential distribution of this duration within the vector population. Our aim was to investigate the validity of this major and extremely common assumption. We developed a stochastic mechanistic compartmental model to represent two successive stages: (1) infection of the mosquito (crossing the intestinal barrier) by an arbovirus; (2) virus dissemination throughout the mosquito body. We infer model parameters using an ABC-SMC (Approximate Bayesian Computation - Sequential Monte Carlo) method, drawing on literature data from experimental infections of different mosquito species (Aedes albopictus and Aedes aegypti) by three major zoonotic arboviruses: dengue, chikungunya, and zika viruses. We demonstrated that the durations of the infected state following exposure to the zika and dengue viruses were clearly not exponentially distributed within the mosquito populations studied, whereas the duration following exposure to the chikungunya virus did not deviate significantly from an exponential distribution. Therefore, the exponential hypothesis cannot be considered universal for representing intra-vector viral infection, potentially calling into question the predictions of population-scale epidemiological models of arboviruses, which are used as a basis for implementing management measures for zoonotic arboviroses.
Spatial spread of infectious diseases with conditional vector preferences
We explore the spatial spread of vector-borne infections with conditional vector preferences, meaning that vectors do not visit hosts at random. Vectors may be differentially attracted toward infected and uninfected hosts depending on whether they carry the pathogen or not. The model is expressed as a system of partial differential equations with vector diffusion. We first study the non-spatial model. We show that conditional vector preferences alone (in the absence of any epidemiological feedback on their population dynamics) may result in bistability between the disease-free equilibrium and an endemic equilibrium. A backward bifurcation may allow the disease to persist even though its basic reproductive number is less than one. Bistability can occur only if both infected and uninfected vectors prefer uninfected hosts. Back to the model with diffusion, we show that bistability in the local dynamics may generate travelling waves with either positive or negative spreading speeds, meaning that the disease either invades or retreats into space. In the monostable case, we show that the disease spreading speed depends on the preference of uninfected vectors for infected hosts, but also on the preference of infected vectors for uninfected hosts under some circumstances (when the spreading speed is not linearly determined). We discuss the implications of our results for vector-borne plant diseases, which are the main source of evidence for conditional vector preferences so far.